Distributions of PM2.5 Source Strengths for ... - ACS Publications

Nov 19, 2005 - Research Triangle Park (RTP) PM panel study. The study (n. ) 37 participants) included monitoring for 7 consecutive days in each of fou...
14 downloads 0 Views 164KB Size
Download PDF
Environ. Sci. Technol. 2006, 40, 163-169

Distributions of PM2.5 Source Strengths for Cooking from the Research Triangle Park Particulate Matter Panel Study DAVID A. OLSON* AND JANET M. BURKE National Exposure Research Laboratory, United States Environmental Protection Agency, Research Triangle Park, North Carolina 27711

Emission rates, decay rates, and cooking durations are reported from continuous PM2.5 (particulate matter less than 2.5 µm) concentrations measured using personal DataRam nephelometers (1-min time resolution) from the Research Triangle Park (RTP) PM panel study. The study (n ) 37 participants) included monitoring for 7 consecutive days in each of four consecutive seasons (summer 2000 through spring 2001). Cooking episodes (n ) 411) were selected using time-activity diaries and criteria for cooking event duration, peak concentration level, and decay curve quality. Averaged across all cooking events, mean source strengths were 36 mg/min (median ) 12 mg/min), mean decay rates were 0.27 h-1 (0.17 h-1), and mean cooking durations were 11 min (7 min). Cooking events were further separated into one of seven categories representing cooking method: burned food (oven cooking, toaster, or stovetop cooking), grilling, microwave, toaster oven, frying, oven cooking, and stovetop cooking. The highest mean source strengths were identified from burned food (mean ) 470 mg/min), grilling (173 mg/min), and frying (60 mg/ min); differences between both burned food and grilling compared with all remaining cooking methods were statistically significant. Source strengths, decay rates, and cooking durations were also compared by season and typical meal times (8:00 a.m., 12:00 p.m., and 6:00 p.m.); differences were generally not statistically significant for these cases. Mean source strengths using electric appliances were typically a factor of 2 greater than those using gas appliances for identical cooking methods (frying, oven cooking, or stovetop cooking), although in all cases the difference was not statistically significant. Distributions of source strengths and decay rates for cooking events were also compared among study subjects to assess both within- and between-subject variability. Each subject’s distribution of source strengths during the study tended to be either lower than the overall study average (and with lower variability) or higher than the overall study average (and with higher variability). No relationships could be found between source strength and either subject characteristics (age, gender, employment status) or home characteristic (daily air exchange rate). The large number of cooking events and the broad range of cooking activities included in this analysis makes the reported distributions of PM2.5 source * Corresponding author phone: (919)541-3190; fax: (919)541-0905; e-mail: [email protected]. 10.1021/es050359t Not subject to U.S. Copyright. Publ. 2006 Am. Chem. Soc. Published on Web 11/19/2005

strengths useful for probabilistic exposure modeling even though the study population was limited.

Introduction Numerous epidemiological and health studies have indicated an association between particulate matter (PM) and human health effects, including premature mortality and morbidity (1-5). Although these studies have typically used fixed-site ambient monitors as a surrogate for PM exposure, human exposure field studies have also been conducted to better characterize actual personal PM exposure (6-9). Characterizing sources of PM in indoor air is of particular interest since people spend most of their time indoors (10). Large-scale studies (more than 150 homes) have identified environmental tobacco smoke (ETS) as a major source of fine and coarse particles (11-14). Several studies have identified ETS and/or cooking as the largest sources of PM indoors (14-19). Other sources of indoor PM include cleaning, such as vacuuming and other activities (17, 20, 21), and resuspension from movement of people (17, 21). One difficulty in characterizing PM indoors is the variable nature of indoor PM sources (19) and house conditions such as ventilation and infiltration (22, 23). Furthermore, other mechanisms governing indoor PM concentrations (deposition and resuspension) have been noted to depend on a number of factors, e.g., particle size (21). Some previous studies (e.g., ref 14) have used sampling times of 12 h or longer, making quantification of emission rates highly uncertain. Recent studies have increasingly used continuous measurements of PM to better characterize PM concentrations indoors (17-19, 24-28). Despite growing scientific understanding of the emissions and prevalence of PM outdoors, few studies have characterized indoor sources of PM directly from field measurements for sources other than environmental tobacco smoke (ETS). Scientific interest in better characterization of emissions from cooking has grown, especially since cookingsparticularly cooking with gas stovesshas been associated with adverse health effects (29-33). Research completed to date measuring PM from cooking generally has involved a limited number of residential homes (n < 5), which complicates drawing general conclusions regarding the nature and extent of emissions. The Research Triangle Park (RTP) PM panel study, conducted by the U.S. EPA’s National Exposure Research Laboratory (NERL), provides a unique data set for characterizing residential indoor PM sources (34). The study included continuous PM2.5 measurements for a cohort of 37 participants from different homes over four consecutive seasons, thus representing a wider range of cooking events and residential conditions. The overall objective of this paper is to provide estimates of parameters needed for characterizing exposure to cooking in residential homes, as well as to assess variability of these parameters. To this end, estimates of PM2.5 emission rates, decay rates, and durations for known cooking events during the study are presented in order to characterize both the overall and within-subject variability.

Methods Study Design. The RTP PM panel study included a cohort of 29 nonsmoking hypertensive subjects and a cohort of eight subjects having implanted cardiac defibrillators (36 residential homes total) living within the RTP area of North Carolina. Participants were monitored for 7 consecutive days VOL. 40, NO. 1, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

163

in each of four consecutive seasons (summer 2000 through spring 2001). Daily (24 h) integrated measurements of PM2.5 (PM less than 2.5 µm) mass, PM2.5 elemental composition, PM10 (PM less than 10 µm) mass, elemental carbon (EC), organic carbon (OC), NO2, O3, and CO were collected concurrently at a centrally located stationary outdoor site (ambient site), outdoors and indoors at each subject’s residence, and on each subject using personal monitors. Air exchange rates were also measured daily (24-h integrated) in each residence using a perfluorotracer method (35). A more detailed description of the study design is given elsewhere (34). In addition to the daily integrated samples, DataRam nephelometers (MIE, pDR-1000, Bedford, MA) were used to collect continuous (1-min averaging time) personal and residential indoor measurements of PM2.5 for each subject (36). As described in more detail elsewhere (36), the DataRam nephelometer collected passive samples and is considered a measure of PM2.5 because the instrument’s highest sensitivity is from 0.5 to 2.0 µm. Although these instruments use optical-based measurements, previous field studies have shown reasonable correlations between personal DataRam nephelometers and PM mass concentrations (25, 26, 3740). For example, Howard-Reed et al. (25) reported an R2 of 0.66 and Liu et al. (39) reported R2 values ranging from 0.77 to 0.84 between personal DataRam nephelometers and 24-h integrated PM2.5 mass concentrations. Researchers have noted an increase in particle readings at relative humidity (RH) greater than 85% (41), although no effect was observed for RH between 20% and 60% typically encountered indoors (25). In addition to continuous PM2.5 measurements, study participants wore a monitor (HOBO, Bourne, MA) which measured relative humidity and temperature every 3 min. Supplementary data were also collected on housing conditions and human activity patterns for each subject. Daily questionnaires and time-activity diaries were collected for each sampling day. Subjects recorded their locations at 15min time intervals among five general categories (inside home, outside near home, inside away from home, outside away from home, and inside vehicle) in their daily activity diary and were instructed to note when certain PM-generating activities were performed (e.g., cooking). Data from the DataRam nephelometers were downloaded daily and also used to verify that descriptive activity information was recorded by each subject for the five highest PM2.5 concentration peaks on each day. Data Analysis. Each subject’s personal DataRam nephelometer data were used to calculate PM2.5 emission rates and decay rates for cooking events. Several criteria were used to assess data quality from the personal DataRam nephelometer (described in more detail in ref 36). The first and last 15 min of each monitoring day were excluded from this analysis. These periods involved set up of instrumentation by field staff and may have led to elevated particle concentrations due to particle resuspension. The possible effect of instrument drift was assessed by measuring zero air with each nephelometer at the end of each monitoring day. Monitoring days where instrument drift exceeded 5 µg/m3 were excluded from this analysis. Data points that were outside the manufacturer’s listed ranged of detection (from 0.001 to 400 mg/m3) were also excluded from this analysis. Because previous researchers noted a bias associated with the DataRam nephelometer measuring in environments with higher humidities (41), all peaks where the average indoor humidity exceeded 75% were removed from this analysis. In addition, all the data for two subjects who were consistently exposed to secondhand ETS in their homes (based on timeactivity diaries) were excluded. For identification of peaks associated with cooking, time periods were restricted to known cooking events based on 164

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 1, 2006

the subject’s daily time-activity diary. Peaks in the personal PM2.5 concentration were then examined, and only those meeting the following criteria were included in the analysis: the increase in concentration was at least 5 min in duration, the peak concentration was at least twice the background concentration, and the concentration returned to background levels. Descriptive activity information recorded in the timeactivity diaries was also examined for all cooking events included in the analysis. Information on the type of cooking (frying, grilling) or appliance (stovetop, oven, microwave, toaster oven) was available for nearly two-thirds of the cooking events. Higher PM2.5 concentration peaks generally had descriptive information in the diary due to the daily verification of the five highest peaks in each subject’s data. It must be noted that the subjects varied in their recording of descriptions and the level of detail provided. Emission rates for cooking were estimated by completing a mass balance for PM2.5 and assuming one well-mixed reactor (zone) for the entire house:

E dC ) λPCout - λC + - kC dt V

(1)

where C is the concentration of PM2.5 in the house (mg/m3), t is time (h), λ is the air exchange rate (h-1), P is the penetration factor (unitless), Cout is the instantaneous concentration of PM2.5 outdoors (mg/m3), E is the emission rate (mg/h), V is the volume of the house (m3), and k is the deposition rate (h-1). Assuming steady-state conditions (dC/dt ) 0) and integrating across the time of the cooking event results in the following expressions:

C(t) ) Cb +

E (1 - e-k′t) k′V

C(t) ) (Cp - Cb) e-k′(t-tp) + Cb

t e tp t g tp

(2) (3)

where Cb is the background concentration of PM2.5 (mg/m3), Cp is the peak concentration (mg/m3), and tp is the time when the peak concentration occurs (h). The first-order decay constant k′ includes removal due to air exchange as well as other processes, e.g., deposition. The decay rate (k′) was calculated for each cooking event using eq 3. The event decay rate was then used in eq 2 along with each subject’s house volume, V, to calculate the emission rate, E, for the cooking event. As part of the RTP panel study, both personal and residential indoor samples were collected using personal nephelometers. Residential indoor samples were collected in the primary living area (nonbedroom area where the participant spent the most time). For three of the study participants, both the personal and residential indoor measurements were examined to assess the difference in emission rates between personal and residential indoor measurements. For these three subjects, estimated emission rates using personal measurements were an average of approximately 9 times (median of 4 times) those estimated using residential indoor measurements. Although these results are expected because the personal measurements were collected closer to the cooking source, the reader should be cautioned regarding these differences. Since the goal of this paper is to estimate PM2.5 emission rates for personal exposures while cooking, only the results of the analysis using the personal nephelometer data are reported. Preparation of raw data, linear regression analyses, descriptive statistics, and nonparametric statistics were all performed using SAS v.8.02 (SAS Institute, Cary, NC). A level of significance of R ) 0.05 was used for all statistical procedures. The Shapiro-Wilk test statistic was used to

FIGURE 1. Example PM2.5 concentration time series and calculated source strength and decay rate for a single cooking event during the RTP PM panel study.

FIGURE 2. Histogram of PM2.5 source strengths for cooking events recorded by subjects in the RTP PM panel study. Figure includes 95% of the data points summarized in Table 1; the remaining 5% (ranging from 110 to 1496 mg/min) were not included to make the graph more visible.

determine whether the data were log-normally distributed. A series of statistical tests (e.g., Wilcoxon scores, KolmogorovSmirnov test, Bonferroni comparisons) were used to determine statistically significant differences in mean values between categories of data (e.g., cooking method). Levene’s test for homogeneity was used to determine statistically significant differences in variances.

regressions for calculating source strengths had an average R2 value of 0.73 and an average p-value of 0.01. The calculated source strengths were log-normally distributed (Figure 2) with a median value of 12 mg/min. Source strengths for all cooking events ranged over 3 orders of magnitude (from 0.6 to 1496 mg/min) with the interquartile range encompassing approximately 1 order of magnitude (from 4.8 to 30 mg/ min). Also shown in Table 1 is the PM2.5 concentration averaged over the cooking events as measured by both the personal and indoor monitors. As expected, in all cases the higher PM2.5 concentration was measured by the personal monitor. This large variability in source strengths across cooking events was examined using the descriptive activity information available from the time-activity diaries as shown in Table 1. Cooking events that included burned food or grilling in their description had the highest mean PM2.5 source strengths (470 ( 530 and 173 ( 92 mg/min, respectively). These events also had the highest median PM2.5 source

Results and Discussion Source Strengths. An example cooking event is shown in Figure 1, where personal PM2.5 concentrations (µg/m3) are displayed as a function of time of day for the time period recorded for cooking in the subject’s time-activity diary (10:30 a.m. to 11:15 a.m.). For this example cooking event, regressions for calculating source strength and decay rate had R2 values greater than 0.85 and were statistically significant (p < 0.001). A statistical summary of PM2.5 source strengths from cooking for all subjects in this study is shown in Table 1. The mean ((standard deviation) source strength for all cooking events was 36 ( 98 mg/min. Across all cooking events,

TABLE 1. Statistical Summary of PM2.5 Source Strengths (mg/min)

all cooking events

percentile distributions of source strengths (mg/min) min p25 median p75 p95 max

PM2.5 concn (mg/m3)d

n

mean

SDa

GMb

GSDc

411

36

98

2.5

1.4

12

30

117

1496

188 (122)

530 92 92 27 35 11 12 63 17

5.7 5.1 3.5 3.8 2.0 2.0 1.8 2.3 2.1

Cooking Method 1.0 116 155 0.5 91 91 1.1 1.9 16 0.5 22 26 1.3 0.6 2.9 0.9 1.4 4.6 1.1 0.9 2.1 1.3 0.8 3.4 1.1 0.7 3.5

231 155 33 45 6.9 7.4 7.7 10 7.9

592 272 71 83 23 15 14 24 18

1496 272 170 89 42 36 32 88 48

1496 272 663 89 219 44 69 360 107

1360 (603) 679 (220) 341 (215) 72 (3) 110 (87) 47 (37) 76 (66) 244 (117) 84 (79)

82 40 104 44 38 10 13 7.8

2.8 2.7 3.6 3.2 2.1 1.6 1.9 1.5

1.4 1.2 1.1 1.0 1.3 1.3 1.1 0.9

Gas vs Electric 0.7 7.6 0.6 4.9 1.9 17 3.8 14 0.7 3.5 0.6 2.5 0.9 2.3 2.0 2.1

18 17 37 27 7.0 4.9 7.8 4.2

123 93 205 103 51 29 32 21

663 233 663 233 219 29 69 21

230 (154) 133 (93) 377 (235) 189 (133) 123 (98) 42 (27) 74 (67) 87 (63)

burned foode grilling frying toaster oven stovetop microwave oven multiple eventsf unspecifiedg

6 3 105 7 40 20 38 43 149

all electric all gas frying-electric frying-gas stovetop-electric stovetop-gas oven-electric oven-gas

141 42 75 30 33 7 33 5

470 173 60 51 17 11 10 29 14 43 29 69 38 19 10 11 6.8

0.6

4.8

46 37 82 43 23 18 14 4.8

a SD ) Standard deviation. b GM ) Geometric mean. c GSD ) Geometric standard deviation. d Average PM 2.5 concentration measured during event, including decay as measured by personal sample (indoor sample in parentheses). e Events using either toaster, oven, or stovetop. f More than one cooking method (e.g., stovetop and oven) specified during occurrence of peak (did not include either burned food or grilling). g No cooking method specified in time-activity diaries.

VOL. 40, NO. 1, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

165

strengths and the highest mean PM2.5 concentrations (1360 and 679 µg/m3, respectively). The number of events with these descriptions was very low (n ) 6 and 3 out of 411 total, respectively), although their impact on PM2.5 exposures can be substantial as indicated by the PM2.5 concentrations listed in Table 1. Differences in source strengths between both burned food and grilling compared with each of the remaining cooking methods listed in Table 1 were statistically significant. Frying had the next highest mean PM2.5 source strength (60 ( 92 mg/min) and mean PM2.5 concentration during the events (341 µg/m3) but made up approximately 25% of all cooking events (n ) 105). Previous research (19) has shown that frying and smoky cooking events are major sources of indoor PM levels, although source strengths were not estimated in that study. The remaining cooking events with information on cooking method had descriptions of the type of appliance used. Toaster oven use had the highest mean and median PM2.5 source strength for these events (51 ( 27 and 45 mg/ min, respectively), but included only seven cooking events. Similar to frying, these combined cooking activities (stovetop, microwave, and oven use) also constituted approximately 25% of all cooking events (n ) 98). Descriptive information for approximately 10% of all cooking events (n ) 43) included use of multiple appliances, while just over one-third (n ) 149) of all cooking events had no descriptive information in the time-activity diary. When the four cooking methods with the lowest mean values (toaster oven, stovetop cooking, microwave, and oven cooking) were combined in one distribution, source strengths for frying were significantly higher than those for the combined distribution. Another noteworthy finding is that for electric and gas appliance usage for comparable cooking methods (frying, oven cooking, and stovetop cooking), mean source strengths were in all cases higher using electric appliances. Although mean source strengths using electric appliances were typically a factor of 2 greater than those using gas appliances, in all cases the difference was not statistically significant. The results do suggest, however, that exposures may be underestimated if gas-only PM2.5 emission rates were extrapolated to a larger population subset, an important consideration given that several previous studies (29-33) have focused only on health effects associating with gas cooking. The trend of higher source strengths using electric appliances may be related to several possible physical explanations. These include pan temperature (stovetop cooking) or oven temperature, type of frying oil used (e.g., using an oil that does not withstand high temperatures), or time for the appliance to achieve its temperature set point. However, another important factor is the effect of outlier source strengths. For example, the electric appliances mean source strength for frying was approximately 80% higher than that for gas appliances. Removal of the three highest electric appliance source strengths (663, 460, and 425 mg/min) results in a mean source strength of 50 mg/min, only about 30% higher than that for gas appliances. One possible explanation for these high frying events is that they also involved burned food but were not identified as such by the study participants. Thus, one consideration for future studies involving PM exposure from cooking is detailed criteria (e.g., pan temperature or food temperature) for descriptive activity information involving events such as burned food. Source strengths were also compared by season and time of day (Table S1 in the Supporting Information). Three time of day categories were defined based on typical meal times (6:00-10:00 a.m., 10:00 a.m.-2:00 p.m., and 4:00-8:00 p.m.) for comparisons of cooking source strengths. There were no statistically significant differences when comparing mean source strengths either among seasons or among time of day categories. In addition, there were no statistically significant 166

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 1, 2006

FIGURE 3. Distributions (box plots) of PM2.5 source strengths (mg/ min) for cooking for all cooking events and by study participant, with the total number of cooking events (black diamonds) for the subject. The box defines the interquartile range and median of the distribution; the whiskers are the 10th and 90th percentiles of the distribution; the dark horizontal bar is the mean value. (a) At least one grilling event. (b) At least 33% of the events were frying. (c) At least one burned food event. (d) Greater than 70% of the events were unspecified. differences when comparing source strengths by start hour (i.e., the hour of day when the cooking event started). Distributions of PM2.5 source strengths were also compared among study subjects to assess both within- and between-subject variability. Figure 3 shows the distribution of PM2.5 source strengths for all cooking events and for 16 subjects with cooking events recorded during all four seasons and at least nine total cooking events during the study. The within-subject variability in PM2.5 source strengths indicates at least three different patterns for the subjects (betweensubject variability) when compared to the overall average for all cooking events. The first type involves subjects with high median values, e.g., approximately 2 or more times the median value for all cooking events (12 mg/min). This group includes subject 17, who had the largest number of cooking events (n ) 32), and had at least 10 events using cooking methods that had high PM2.5 source strengths (all three of the grilling events and all seven of the toaster oven events were by this subject). Several other subjects with high median values in Figure 3 were categorized as frying (i.e., subjects where at least 33% of their cooking events involved frying). Not all subjects categorized under frying had high median values (subjects 4, 14, and 19), a result that is expected given the range of source strengths for this cooking method category (from 2 to 663 mg/min). Many of the lower values in the frying distribution were included with these subjects. The next type of source strength distribution involves subjects where their median value was similar to the median for all events, but with higher mean and 95th percentiles than other subjects. These included all subjects having one or more burned food events, resulting in a highly skewed distribution for those subjects. Even for subjects with a comparatively high number of cooking events (e.g., subject 24 with n ) 30), source strengths distributions were clearly influenced by these outlier values. The next type of source strength distribution involves subjects where their median value was below the median for all events. These included subjects where a majority of their cooking events (>70%) did not have descriptive information. Although all of these subjects had comparatively narrow source strength distributions (possibly indicating recurring

activity patterns), more descriptive information would be needed to evaluate this pattern. The observed difference in source strength variability between subjects was not related to the number of cooking events for each subject (Figure 3). Other known factors were explored, including gender, age, employment status, daily air exchange rate, and stove type, but no relationships were found that further explained differences in PM2.5 source strength variability among subjects. Cooking Durations and Decay Rates. The duration of PM2.5 emissions during cooking events was also examined. Across all cooking events, the mean ((SD) source duration (time between baseline and peak concentration) was 11 ( 14 min (see Table S2 in the Supporting Information). In general, differences among cooking methods, seasons, and time of day were not statistically significant, although the fall and summer seasons and the “dinner” and “breakfast” periods were both statistically different from each other. The source duration data were log-normally distributed with a median value of 7.0 min, and most events (>95%) produced emissions for less than 30 min. The short emission duration relative to typical PM2.5 sampling periods of 12 or 24 h underscores the importance of finer time resolution of sample collection when transient sources such as cooking are present (if identifying and quantifying those cooking sources are desired). The calculation of PM2.5 emission rates for cooking required the estimation of event decay rates. The decay rates reported for this study incorporate all PM2.5 removal mechanisms including air exchange rate. Across all cooking events, the mean decay rate was 0.27 ( 0.30 h-1 (Table S3 in the Supporting Information). The decay rates were log-normally distributed with a median value of 0.17 h-1, and the majority of decay rates for cooking events were between 0.10 and 0.40 h-1. Within-subject variability in decay rates was more consistent across subjects than were source strengths (Figure S1 in the Supporting Information). This result is expected as each subject’s home ventilation habits are assumed to be consistent across the study period. All the subjects had mean decay rates between 0.07 and 0.56 h-1, with interquartile ranges between 0.04 and 0.67 h-1. There was no relationship between subjects with higher variability in source strengths and either higher or lower variability in decay rates. There was also no correlation between variability in daily air exchange rate and variability in decay rates for each subject. However, this comparison is complicated by the fact that air exchange rates were measured over a 24-h sampling time while cooking events for this study were completed within minutes. Comparison of Results to Those of Previous Studies. O ¨ zkaynak et al. (14) reported PM2.5 emission rates for cooking estimated using a nonlinear least-squares solution to the mass-balance equation for PTEAM study data. The mean emission rate from cooking was 1.7 ( 0.6 mg/min using 12-h integrated gravimetric PM2.5 samples collected at an indoor location in the home, compared to 36 ( 98 mg/min estimated from this study using the personal DataRam nephelometers. The PTEAM study included a larger number of homes (n ) 178), but the estimated emission rates likely underestimated the actual emission rates for study subjects since only stationary indoor samples integrated over 12 h were collected for PM2.5. One other previous study that estimated PM source strengths for cooking involved particle size distribution measurements in a small number of residential homes. Abt et al. (17, 18) collected 5-min measurements inside and outside of four single-family homes for either one or two 6-day sampling periods using a scanning mobility particle sizer (SMPS) and aerodynamic particle sizer (APS). Emission

rates of PM(0.7-10) based on a mass-balance model were 15.94 ( 6.77 µm3/cm3/h for cooking (18). The use of a specific gravity of 1.0 for combustion particles (36) implies that a particle volume of 1 µm3/cm3 corresponds to a mass of 1 µg/m3. The corresponding emission rates for PM(0.7-10) are then 15.94 ( 6.77 µg/m3/h. A larger number of studies have involved estimation of decay rates. Reported decay rates from selected studies are as follows: 0.05 ( 0.05 h-1 measured inside five museums (42); 0.3 ( 0.1 h-1 measured from telephone switching stations (43, 44); 0.39 ( 0.16 h-1 estimated from 12-h samples from 178 study participants (14); ranging from 0.46 to 1.07 PM(1-5) based on samples from one house in summer (21); 0.50 ( 0.11 h-1 based on four cooking events (45); 0.7 ( 0.2 h-1 measured from nine homes in Boston (27); and 1.18 ( 0.25 h-1 measured from four homes in Boston (18). The mean decay rate from this study (0.27 ( 0.30 h-1) is similar in magnitude but somewhat lower than the mean values reported in the previous studies. Recommendations and Limitations. These results provide a more comprehensive analysis of PM2.5 emission rates from cooking than has been possible with previous studies. The 1-min PM2.5 concentration data from personal DataRam nephelometers for 35 different subjects over a 1 week period for each of four seasons provided over 400 cooking events for the analysis. The majority of cooking events lasted less than 10 min, indicating that time-resolved measurements and detailed activity diaries are needed to accurately quantify the impact of cooking events on personal PM2.5 exposures. The results also demonstrate that PM2.5 emission rates from cooking span orders of magnitude, and as such a large number of cooking events is needed to adequately characterize PM2.5 source strengths for cooking. Although the overall distributions of PM2.5 source strengths likely include a broad range of cooking activities, the frequency of high PM2.5generating cooking activities strongly influenced each subject’s distribution. The distributions of PM2.5 source strengths estimated from this study are useful for probabilistic exposure modeling where exposures are estimated for individuals by simulating daily activity patterns and the concentrations within the locations people spend time in (46). Since people spend 69% of their day in a residence on average (47), where outdoor concentrations of PM2.5 combine with PM2.5 emitted from indoor sources, distributions of PM2.5 source strengths for cooking and other indoor sources such as cleaning are needed to improve estimates of exposure using probabilistic models. The overall distribution for all cooking events provides an improved distribution for use in exposure modeling that characterizes variability across the range of different types of cooking performed by the subjects in this study. For exposure models with the capability to separate homes by the type of stove present, the reported distributions for homes with gas and electric stoves could be used. The type of cooking method appears to be an important factor influencing PM2.5 emissions, although the number of cooking events in this analysis was not sufficient to develop separate distributions for each cooking method. The differences observed among subjects in source strength variability for cooking suggests that exposure models should include two cooking categories for simulated individuals: individuals with low PM2.5generating cooking activities and individuals with highly variable cooking source strengths, that can be randomly assigned without dependence on the individual’s demographic characteristics. The patterns comparing withinsubject variability suggest the prevalence of recurring activity patterns and that certain cooking activities (e.g., grilling, burned food, and frying) are especially influential on a subject’s distribution of source strengths. VOL. 40, NO. 1, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

167

While the RTP PM panel study included a relatively large number of subjects for a panel study and collected longitudinal measurements over four seasons, the data and resulting distributions have limitations. Several studies have reported that personal DataRam nephelometers overestimate gravimetric concentrations (19, 25, 48-50), typically between 30 and 50%. Observed slopes between nephelometer and gravimetric data were 1.5 for personal PM measurements and 1.7 for indoor PM measurements for this study (36). The distributions may not be representative of a broader population, since the subjects were from a relatively small geographic area in North Carolina, were not selected randomly or as a probability sample, and were all over 55 years of age. In addition, estimation of PM2.5 source strengths for cooking was not a primary goal of the RTP PM panel study, and therefore, detailed information on each cooking event (e.g., cooking temperature) was limited. Future studies attempting to understand PM2.5 emissions from cooking and its impact on personal PM2.5 exposures should collect this important human activity information.

(8)

(9)

(10)

(11)

(12)

Acknowledgments The authors thank Anne Rea, Carry Croghan, and Lance Wallace of the U.S. Environmental Protection Agency for creating the combined human activity and PM2.5 measurement database used in this analysis. Considerable effort was required to produce a complete and quality assured data set usable for data analyses such as what has been presented in this paper. The authors also acknowledge the efforts of those responsible for planning and conducting the RTP PM panel study, including Ron Williams of the U.S. Environmental Protection Agency and Charles Rodes of RTI International, as well as all the study participants who completed the monitoring and recording of activities. Disclaimer: The United States Environmental Protection Agency through its Office of Research and Development funded and managed the research described here. It has been subjected to Agency’s administrative review and approved for publication as an EPA document. Mention of trade names or commercial products does not constitute endorsement or recommendation for use.

(13)

(14)

(15) (16)

(17)

(18)

(19)

Supporting Information Available Information on summary statistics from source strengths, source durations, and decay rates, and within-subject variability of decay rates. This material is available free of charge via the Internet at http://pubs.acs.org.

(20)

(21)

Literature Cited (1) Dockery, D. W.; Pope, C. A.; Xu, X.; Spengler, J. D.; Ware, J. H.; Fay, M. E.; Ferris, B. G.; Speizer, F. E. An association between air pollution and mortality in six U. S. cities. New England Journal of Medicine 1993, 329, 1753-1759. (2) Dockery, D. W.; Pope, C. A. Acute respiratory effects of particulate air pollution. Annu. Rev. Public Health 1994, 15, 107-132. (3) Pope, C. A.; Bates, D.; Raizenne, M. Health effects of particulate air pollution-time for reassessment. Environ. Health Perspect. 1995, 103, 472-480. (4) Schwartz, J.; Dockery, D. W.; Neas, L. M. Is daily mortality associated specifically with fine particles? J. Air Waste Manage. Assoc. 1996, 46, 927-939. (5) Samet, J.; Dominici, F.; Curriero, F.; Coursac, I.; Zeger, S. Fine particulate air pollution and mortality in 20 U. S. cities, 19871994. New England Journal of Medicine 2000, 343, 1742-1749. (6) Clayton, C. A.; Perritt, R.; Pellizzari, E.; Thomas, K.; Whitmore, R.; Wallace, L.; O ¨ zkaynak, H.; Spengler, J. Particle Total Exposure Assessment Methodology (PTEAM): study: Distributions of aerosol and elemental concentrations in personal, indoor, and outdoor air samples in a southern California community. J. Exposure Anal. Environ. Epidemiol. 1993, 3, 227-250. (7) Jantunen, N.; Hanninen, O.; Katsouyanni, K.; Knoppel, H.; Kunzli, N.; Lebret, E.; Maroni, M.; Saarela, K.; Sram, R.; Zmirou, D. Air 168

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 1, 2006

(22)

(23)

(24)

(25)

(26)

pollution exposure in European cities: The “EXPOLIS” study. J. Exposure Anal. Environ. Epidemiol. 1998, 8, 495-518. Williams, R.; Suggs, J.; Zweidinger, R.; Evans, G.; Creason, J.; Kwok, R.; Rodes, C.; Lawless, P.; Sheldon, L. The 1998 Baltimore particulate matter epidemiology-exposure study: Part 1-comparison of ambient, residential outdoor, indoor and apartment particulate matter monitoring. J. Exposure Anal. Environ. Epidemiol. 2000, 10, 518-532. Williams, R.; Suggs, J.; Creason, J.; Rodes, C.; Lawless, P.; Kwok, R.; Zweidinger, R.; Sheldon, L. The 1998 Baltimore particulate matter epidemiology-exposure study: Part 2-personal exposure assessment associated with an elderly study population. J. Exposure Anal. Environ. Epidemiol. 2000, 10, 533-543. Klepeis, N. E.; Tsang, A. M.; Behar, J. V. Analysis of the National Human Activity Pattern Survey (NHAPS) Respondents from a Standpoint of Exposure Assessment; EPA/600/R-96/074; U.S. Environmental Protection Agency, Office of Research and Development, Washington, D.C., 1996. Spengler, J. D.; Dockery, D. W.; Turner, W. A.; Wolfson, J. M.; Ferris, B. G. Long-term measurements of respirable sulfates and particles inside and outside homes. Atmos. Environ. 1981, 15, 23-30. Neas, L. M.; Dockery, D. W.; Ware, J. H.; Spengler, J. D.; Ferris, B. G.; Speizer, F. E. Concentration of indoor particulate matter as a determinant of respiratory health in children. Am. J. Epidemiol. 1994, 139, 1088-1099. Koutrakis, P.; Briggs, S. L. K.; Leaderer, B. P. Source apportionment of indoor aerosols in Sulfolk and Onondaga counties, New York. Environ. Sci. Technol. 1992, 26, 521-527. O ¨ zkaynak, H.; Xue, J.; Spengler, J.; Wallace, L.; Pellizzari, E.; Jenkins, P. Personal exposure to airborne particles and metals: Results from the Particle TEAM study in Riverside, California J. Exposure Anal. Environ. Epidemiol. 1996, 6, 57-78. Wallace, L. Indoor particles: A review. J. Air Waste Manage. Assoc. 1996, 46, 98-126. Janssen, N. A. H.; Hoek, G.; Brunekreef, B.; Harssema, H.; Mensink, I.; Zuidhof, A. Personal sampling of particles in adults: Relation among personal, indoor, and outdoor air concentrations. Am. J. Epidemiol. 1998, 147, 537-547. Abt, E.; Suh, H. H.; Allen, G.; Koutrakis, P. Characterization of indoor particle sources: A study conducted in the metropolitan Boston area. Environ. Health Perspect. 2000, 108, 35-44. Abt, E.; Suh, H. H.; Allen, G.; Koutrakis, P. Relative contribution of outdoor and indoor particle sources to indoor concentrations Environ. Sci. Technol. 2000, 34, 3579-3587. Wallace, L. A.; Mitchell, H.; O’Connor, G. T.; Neas, L.; Lippmann, M.; Kattan, M.; Koenig, J.; Stout, J. W.; Vaughn, B. J.; Wallace, D.; Walter, M.; Adams, K.; Liu, L.-J. S. Particle concentrations in inner-city homes of children with asthma: The effect of smoking, cooking, and outdoor pollution. Environ. Health Perspect. 2003, 111, 1265-1272. Kamens, R.; Lee, C.-T.; Wiener, R.; Leith, D. A study to characterize indoor particles in three nonsmoking homes. Atmos. Environ. A 1991, 25, 939-948. Thatcher, T. L.; Layton, D. W. Deposition, resuspension, and penetration of particles within a residence. Atmos. Environ. 1995, 29, 1487-1497. Sarnet, J. A.; Koutrakis, P.; Suh, H. H. Assessing the relationship between personal particulate and gaseous exposures of senior citizens living in Baltimore, MD. J. Air Waste Manage. Assoc. 2000, 50, 1184-1198. Wallace, L. A.; Howard-Reed, C. H.; Emmerich, S. J. Continuous monitoring of ultrafine, fine, and coarse particles in a residence for 18 months in 1999-2000. J. Air Waste Manage. Assoc. 2002, 52, 174-190. Brauer, M.; Hirtle, R.; Lang, B.; Ott, W. Assessment of indoor fine aerosol contributions from environmental tobacco and cooking with a portable nephelometer. J. Exposure Anal. Environ. Epidemiol. 2000, 10, 136-144. Howard-Reed, C.; Rea, A. W.; Zufall, M. J.; Burke, J. M.; Williams, R. W.; Suggs, J. C.; Sheldon, L. S.; Walsh, D.; Kwok, R. Use of a continuous nephelometer to measure personal exposure to particles during the U.S. Environmental Protection Agency Baltimore and Fresno panel studies. J. Air Waste Manage. Assoc. 2000, 50, 1125-1132. Rea, A. W.; Zufall, M. J.; Williams, R. W.; Howard-Reed, C.; Sheldon, L. The influence of human activity patterns on personal PM exposure: A comparative analysis of filter-based and continuous particle measurements. J. Air Waste Manage. Assoc. 2001, 51, 1271-1279.

(27) Long, C. M.; Suh, H. H.; Catalano, P. J.; Koutrakis, P. Using timeand size-resolved particulate data to quantify indoor penetration and deposition behavior. Environ. Sci. Technol. 2001, 35, 20892099. (28) Wallace, L. A.; Emmerich, S. J.; Howard-Reed, C. Source strengths of ultrafine and fine particles due to cooking with a gas stove. Environ. Sci. Technol. 2004, 38, 2304-2311. (29) Goldstein, B. D.; Melia, R. J.; Chinn, S.; Florey, C. V.; Clark, D.; John, H. H. The relation between respiratory illness in primary schoolchildren and the use of gas for cooking-II. Factors affecting nitrogen dioxide levels in the home. Int. J. Epidemiol. 1979, 8, 339-345. (30) Seaton, A.; Dennekamp, M. Hypothesis: Ill health associated with low concentrations of nitrogen dioxide-an effect of ultrafine particles? Thorax 2003, 58, 1012-1015. (31) Volkmer, R. E.; Ruffin, R.; Wigg, N. R.; Davies, N. The prevalence of respiratory symptoms in south Australian preschool children. 2. Factors associated with indoor air quality. Journal of Pediatrics and Child Health 1995, 31, 116-120. (32) Jarvis, D.; Chinn, S.; Luczynska, C.; Burney, P. Association of respiratory symptoms and lung function in young adults with use of domestic gas appliances. Lancet 1996, 347, 426-431. (33) Jarvis, D.; Chinn, S.; Sterne, J.; Luczynska, C.; Burney, P. The association of respiratory symptoms and lung function with the use of gas for cooking. Eur. Respir. J. 1998, 11, 651-658. (34) Williams, R.; Suggs, J.; Rea, A.; Leovic, K.; Vette, A.; Croghan, C.; Sheldon, L.; Rodes, C.; Thornburg, J.; Ejire, A.; Herbst, M.; Sanders, W. The Research Triangle Park particulate matter panel study: PM mass concentration relationships. Atmos. Environ. 2003, 37, 5349-5363. (35) Dietz, R.; Cote, E. Air infiltration measurements in a home using a convenient perfluorocarbon tracer technique. Environ. Int. 1982, 8, 419-433. (36) Wallace, L.; Williams, R.; Rea, A.; Croghan, C. Continuous weeklong measurements of personal exposures and indoor concentrations of fine particles for 37 health-impaired North Carolina residents for up to four seasons. Atmos. Environ., in press. (37) Sioutas, C.; Kim, S.; Chang, M.; Terrell, L. L.; Gong, H. Field evaluation of a modified DataRAM MIE scattering monitor for real-time PM2.5 mass concentration measurements Atmos. Environ. 2000, 34, 4829-4838. (38) Quintana, P. J. E.; Valenzia, J. R.; Delfino, R. J.; Liu, L.-J. S. Monitoring of 1-min personal particulate matter exposures in relation to voice-recorded time-activity data. Environ. Res., A 2001, 87, 199-213. (39) Liu, L.-J. S.; Slaughter, J. C.; Larson, T. V. Comparison of light scattering devices and impactors for particulate measurements in indoor, outdoor, and personal environments. Environ. Sci. Technol. 2002, 36, 2977-2986. (40) Chakrabarti, B.; Fine, P.; Defino, R.; Sioutas, C. Performance evaluation of the active-flow personal DataRAM PM2.5 mass

(41)

(42) (43)

(44)

(45)

(46)

(47)

(48)

(49)

(50)

monitor (Thermo Anderson pDR-1200) designed for continuous personal exposure measurements. Atmos. Environ. 2004, 38, 3329-3340. Quintana, P. J. E.; Samimi, B. S.; Kleinman, M. T.; Liu, S. L.; Soto, K.; Warner, G. Y.; Bufalino, C.; Valencia, J.; Francis, D.; Hovell, M. H.; Delfino, R. J. Evaluation of a real-time passive personal particle monitor in fixed site residential indoor and ambient measurements. J. Exposure Anal. Environ. Epidemiol. 2000, 10, 437-445. Ligocki, M. P.; Liu, H. I. H.; Cass, G. R.; John, W. Measurements of particle deposition rates inside southern California museums. Aerosol Sci. Technol. 1990, 13, 85-101. Sinclair, J. D.; Psota-Kelty, L. A.; Weschler, C. J.; Shields, H. C. Measurement and modeling of airborne concentrations and indoor surface accumulation rates of ionic substances at Neenah, Wisconsin. Atmos. Environ., Part A 1990, 24, 627-638. Sinclair, J. D.; Psota-Kelty, L. A.; Peins, G. A.; Ibidunni, A. O. Indoor/outdoor relationships of airborne ionic substances: Comparison of electronic equipment room and factory environments. Atmos. Environ., Part A 1992, 26, 871-882. Howard-Reed, C.; Wallace, L. A.; Emmerich, S. J. Effect of ventilation systems and air filters on decay rates of particles produced by indoor sources in an occupied townhouse. Atmos. Environ. 2003, 37, 5295-5306. Burke, J. M.; Zufall, M. J.; O ¨ zkaynak, H. A population exposure model for particulate matter: Case study results for PM2.5 in Philadelphia, PA. J. Exposure Anal. Environ. Epidemiol. 2001, 11, 470-489. Klepeis, N. E.; Nelson, W. C.; Ott, W. R.; Robinson, J. P.; Tsang, A. M.; Switzer, P.; Behar, J. V.; Hern, S. C.; Engleman, W. H. The National Human Activity Pattern Survey (NHAPS): A resource for assessing exposure to environmental pollutants. J. Exposure Anal. Environ. Epidemiol. 2001, 11, 231-252. Liu, L.-J. S.; Box, M.; Kalman, D.; Kaufman, J.; Koenig, J.; Larson, T.; Sheppard, L.; Slaughter, C.; Lewtas, J.; Wallace, L. A. Exposure assessment of particulate matter for susceptible populations in Seattle, WA. Environ. Health Perspect. 2003, 111, 909-918. Allen, R.; Larson, T.; Sheppard, L.; Wallace, L.; Liu, L.-J. S. Use of Real-time Light Scattering data to estimate the contribution of infiltrated and indoor-generated particles to indoor air. Environ. Sci. Technol. 2003, 37, 3484-3492. Allen, R.; Larson, T.; Sheppard, L.; Wallace, L.; Liu, L.-J. S. Estimated hourly personal exposures to ambient and nonambient particulate matter among sensitive populations in Seattle, Washington. J. Air Waste Manage. Assoc. 2004, 54, 1197-1211.

Received for review February 22, 2005. Revised manuscript received August 5, 2005. Accepted October 11, 2005. ES050359T

VOL. 40, NO. 1, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

169