Ultrafine Particles: Exposure and Source Apportionment in 56 Danish

Article pubs.acs.org/est

Ultrafine Particles: Exposure and Source Apportionment in 56 Danish Homes Gabriel Bekö,*,† Charles J. Weschler,†,‡ Aneta Wierzbicka,§ Dorina Gabriela Karottki,∥ Jørn Toftum,† Steffen Loft,∥ and Geo Clausen† †

International Centre for Indoor Environment and Energy, Dept. of Civil Engineering, Technical University of Denmark, Nils Koppels Allé 402, 2800-Lyngby, Denmark ‡ Environmental and Occupational Health Sciences Institute, University of Medicine and Dentistry of New Jersey and Rutgers University, 170 Frelinghuysen Road, Piscataway, New Jersey 08854, United States § Division of Ergonomics and Aerosol Technology, Lund University, P.O. Box 118, SE-221 00 Lund, Sweden ∥ Department of Public Health, University of Copenhagen, O̷ ster Farimagsgade 5, 1014 Copenhagen, Denmark S Supporting Information *

ABSTRACT: Particle number (PN) concentrations (10−300 nm in size) were continuously measured over a period of ∼45 h in 56 residences of nonsmokers in Copenhagen, Denmark. The highest concentrations were measured when occupants were present and awake (geometric mean, GM: 22.3 × 103 cm−3), the lowest when the homes were vacant (GM: 6.1 × 103 cm−3) or the occupants were asleep (GM: 5.1 × 103 cm−3). Diary entries regarding occupancy and particle related activities were used to identify source events and apportion the daily integrated exposure among sources. Source events clearly resulted in increased PN concentrations and decreased average particle diameter. For a given event, elevated particle concentrations persisted for several hours after the emission of fresh particles ceased. The residential daily integrated PN exposure in the 56 homes ranged between 37 × 103 and 6.0 × 106 particles per cm3·h/day (GM: 3.3 × 105 cm−3·h/day). On average, ∼90% of this exposure occurred outside of the period from midnight to 6 a.m. Source events, especially candle burning, cooking, toasting, and unknown activities, were responsible on average for ∼65% of the residential integrated exposure (51% without the unknown activities). Candle burning occurred in half of the homes where, on average, it was responsible for almost 60% of the integrated exposure.

■

INTRODUCTION

between ozone and terpenes can also generate UFP indoors.18,19 Bhangar et al.,20 Mullen et al.,21,22 and Wallace and Ott23 present a useful metric to describe indoor exposure to UFP. They report the daily integrated particle number (PN) exposure in units of particles per cm3·h/day. The daily integrated exposure is a normalized form of integrated exposure. Since people spend significant time in their homes, where frequent indoor particle source events result in higher PN concentrations, a large fraction of the total ultrafine particle exposure occurs in the home.24 Wallace and Ott23 estimated that for a typical suburban lifetime in the U.S., 47% of the average daily UFP exposure was due to indoor sources, 36% to outdoor sources, and 17% to in-vehicle exposure. Mullen et al.22 concluded that the daily integrated exposure to PN

A review panel assembled by the U.S. Health Effects Institute recently concluded that experimental and epidemiologic studies provide suggestive, but not consistent, evidence of adverse effects of short-term exposures to ambient ultrafine particles (UFP; particles smaller than 100 nm), while information on long-term exposure is not yet available.1 The literature indicating that ultrafine particles may adversely affect human health includes reviews by Delfino et al.,2 Weichenthal et al.,3 Mills et al.,4 and Rückerl et al.5 Compared with larger particles, UFP have higher deposition rates in the lower respiratory tract.6,7 Ultrafine particles enter the indoor environment from outdoors.8,9 However, given their larger diffusivities,6 a smaller fraction of UFP penetrate buildings than do 0.1−0.5 μm diameter particles.10,11 UFP also originate within the indoor environment. Major indoor sources include cooking, tobacco smoking, candle and incense burning, and the use of gas and electric appliances.12−17 Chemical reactions such as those © 2013 American Chemical Society

Received: Revised: Accepted: Published: 10240

June 1, 2013 July 28, 2013 August 19, 2013 August 19, 2013 dx.doi.org/10.1021/es402429h | Environ. Sci. Technol. 2013, 47, 10240−10248

Environmental Science & Technology

Article

the bathrooms of apartments (25% of the investigated homes). All homes were heated using water radiators or floor heating. None of the homes used a gas or oil furnace or water heater. According to the questionnaire filled in by the occupants, 23 homes were located at a distance less than 50 m from a heavily trafficked roadway, 25 homes within 50−250 m, and eight homes at more than 250 m from a roadway. Thirty-three homes were equipped with electric stovetops, while 23 homes used gas stovetops without a pilot light. Only two homes were equipped with a gas oven; the rest used electric ovens. Instrumentation and Data Collection. PN concentrations in the size range between 10 and 300 nm and numberaveraged particle diameters were measured with a NanoTracer PNT1000 (Philips Aerasense, Eindhoven, Netherlands). Three identical, new instruments were purchased prior to the commencement of the measurements. The accuracy of the measured concentrations and particle sizes is ±30%. Additional information regarding the instruments’ operation, accuracy, and how they compared to more advanced particle counters is presented in the Supporting Information. Since the instrument counted particles as large as 300 nm, and UFP are defined as being smaller than 100 nm, the number concentrations reported in this paper are not strictly UFP concentrations. However, particles larger than 100 nm generally contribute little to the total PN concentration12,14 (with the exception of certain types of candles and burning modes31,32) and the impact of indoor sources is especially pronounced in the ultrafine mode.13,17,33,34 Hence, the data obtained from the instruments are considered to be a reasonable proxy for ultrafine particle levels. In each home, one instrument was placed at a height between 0.5 and 1.5 m above floor level in the living room. The measurements always began on a Monday or Wednesday and lasted until Wednesday or Friday, respectively; the duration of monitoring was ∼45 h in each home except for one, where the instrument stopped recording data after about 13 h. The instrument operated continuously with a 16-s sampling interval. Simultaneous PN concentrations were not measured outdoors. The measurements were performed between late October 2011 and mid-February 2012. Valid results were obtained for 56 homes. During the measurement period, the occupants were asked to maintain their normal behavior patterns and to record these in an activity log book. This focused on activities and observations that may have been related to the UFP levels in the homes. Questions were asked about the sensation of tobacco smoke from neighbors or smoking guests, periods of the day when the home was vacant/occupied, and the exact time periods when various activities took place or appliances were in operation. These specifically included boiling, frying, baking, oven operation, stovetop operation, toasting, microwave use, kitchen exhaust use, vacuuming, sweeping, candle burning, incense burning, air freshener use, washing machine use, tumbler drier use, ironing, printer operation, making the bed, and opening windows. Data Analysis. Average indoor PN concentrations and particle diameters were calculated for (1) the full monitoring period as well as for hours when (2) the home was occupied by at least one of the residents and (3) the home was unoccupied. We also analyzed the data from the occupied periods separately for periods between midnight and 06:00 a.m. when the occupants were assumed to be asleep (unless the diaries indicated that the home was unoccupied during part of this

concentrations in California classrooms was about one-sixth of that in homes investigated in the same geographic area. In a personal monitoring study in Copenhagen, 90 min of biking in heavy traffic accounted for only 20% of the integrated PN exposure.25 Bhangar et al.20 apportioned each occupant’s daily integrated residential UFP exposure into contributions from outdoors, episodic indoor sources, and continuous indoor sources. The mean contribution of indoor episodic sources to the daily UFP exposure in these seven northern California residences was ∼59%. In a similar study, the fraction of exposure attributable to indoor sources was smaller in Beijing, reflecting extremely high outdoor PN concentrations and frequent use of natural ventilation.21 Indoor sources were responsible for 50−80% of all UFP measured over more than a year in an occupied home.13,26 Median estimates for the contribution of indoor sources to indoor particle levels were between 58% and 69% in 45 homes of nonsmoking adults and 49 homes of asthmatic children in Canada.27 In contrast to residential environments, Mullen et al.22 found that particles of outdoor origin were the major contributors to the overall indoor PN levels in classrooms. Most of the studies cited above were conducted in a limited number of homes or classrooms. Newly developed portable instruments make it easier to conduct continuous measurements in a larger number of residences. The Center for Indoor Air and Health in Dwellings (CISBO) is a cross-disciplinary research consortium with a goal to develop a scientific basis for improving the built environment and to promote healthy buildings, with a special focus on private housing.28 As part of the study, the association between particle concentrations in dwellings and pulmonary and cardiovascular health indicators was studied among middle-aged and elderly occupants of 56 residences in Copenhagen, Denmark. In this paper, we focus on the measurements of UFP, which were performed continuously over a period of ∼2 days in each home. This paper aims to (1) characterize the indoor exposures of the residents to UFP using a normalized metric for integrated personal exposure assessment, (2) identify the indoor activities responsible for the exposures, and (3) estimate the contribution of various source events to the total exposures of the occupants. Thus, the paper addresses several areas that have received limited attention and which constitute major challenges toward the goal of fully understanding and quantifying human exposure to ultrafine particles.29

■

METHODS Description of Homes. Residents of 58 homes were recruited for a cross-sectional study that involved home inspections, measurements of indoor environmental parameters, and measurements of a number of health indicators. The selected participants were healthy nonsmokers, 41 years or older, living in Copenhagen at least the last 6 months, preferably near a roadway. All participants were recruited randomly (among the nonsmokers) from a cohort of middleaged citizens of Copenhagen.30 The study was approved by the regional ethics committee (Case No. H-4-2010-102). All homes were located within 5 km of the city center. Fortyone of the homes were apartments, while 15 were family houses. The average floor area of the homes was 125 m2, and between 1 and 4 people (on average 2 persons) lived in each home. The homes in Denmark are almost exclusively naturally ventilated, occasionally with additional exhaust ventilation in 10241

dx.doi.org/10.1021/es402429h | Environ. Sci. Technol. 2013, 47, 10240−10248


Article

Table 1. Summary of PN Concentrations, Particle Diameters, and Durations for Time Periods Defined by Occupancya variable

mean (SD)

PN, average in each home (103 cm−3) PN, 95th %ile in each home (103 cm−3) diameter, average in each home (nm) duration (h)

29.1 (40.2) 151.2 (248.3) 76 (14) 45.3 (4.8)

PN, average in each home (103 cm−3) PN, 95th %ile in each home (103 cm−3) diameter, average in each home (nm) duration (h/d)

35.5 (52.3) 177.1 (284.1) 77 (14) 19.5 (3.1)


47.8 (69.5) 221.3 (357.8) 70 (13) 13.5 (2.9)


10.4 (27.3) 38.4 (142.4) 90 (18) 6.0 (0.7)


8.2 (9.9) 16.2 (18.8) 73 (15) 4.8 (3.0)

geometric mean (GSD) total period 15.6 (3.0) 56.7 (4.2) 75 (1.2) 44.9 (1.2) total occupied periodb 17.3 (3.3) 63.3 (4.4) 75 (1.2) 19.3 (1.2) occupants awake 22.3 (3.4) 78.0 (4.4) 69 (1.2) 13.2 (1.3) occupants asleep 5.1 (2.6) 10.4 (3.4) 89 (1.2) 6.0 (1.2) unoccupied periodc 6.1 (2.0) 11.2 (2.2) 72 (1.2) 3.6 (2.5)

median

95th percentile

max

12.4 41.3 76 46.3

106.0 877 102 48.0

251.0 1185 114 48.5

13.1 41.9 76 19.8

144.8 961 103 24.0

304.4 1269 112 24.0

16.9 51.4 69 13.9

215.7 1256 95 17.7

348.5 1594 101 17.9

4.0 7.7 90 6.2

34.2 129.6 126 6.6

201.7 1033 130 6.8

6.2 11.0 75 4.8

20.5 71.6 104 10.7

71.2 101.6 116 12.1

a

The tabulated values for each variable were calculated from single values of the respective variable obtained for each home (n = 56). For example, the first value of 29.1 × 103 cm−3 is the mean PN concentration calculated from the 56 average PN concentrations, each determined from the raw data obtained every 16 s during the entire measurement period. All differences between PN concentrations during the various occupancy periods were statistically significant; p < 0.05. bBased on all occupied periodsperiods when occupants were awake or asleep. cBased on 52 homes only. Four homes were occupied throughout the entire measurement period.

period) and periods when they were awake (rest of the occupied period). The integrated exposures resulting from residential indoor exposures during periods when the homes were occupied were calculated as described by Bhangar et al.20 and Mullen et al.,21 i.e., by integrating the concentration over time (units: particles per cm3·h). The average residential daily integrated exposure was calculated by dividing the integrated exposure by the monitoring duration in days and has units of particles per cm3·h/day (cm−3 h/d throughout the manuscript). Our exposures correspond to the exposure of an occupant who would occupy the home during the entire period indicated as “occupied” by at least one person. The residential integrated UFP exposure during occupied periods in each home was apportioned into contributions from indoor background UFP level and from source events. The average indoor background PN concentration in each home was calculated as the time-weighted average of the concentration during the periods when the occupants were asleep and when the home was unoccupied. This way, both daytime and nighttime background concentrations were considered. Elevated particle concentrations following an identified source event that may have occurred shortly before the beginning of the selected time periods of no occupancy and sleep were excluded. Thus, mainly periods of relatively stable concentrations were used. We assume that the average indoor background PN concentrations obtained in this way reasonably represent the indoor PN levels attributable to the penetration of outdoor particles. This is supported by reasonable agreement

between average indoor background concentrations and the indoor proportions of outdoor particles. The latter were estimated for 29 homes using average outdoor PN concentrations in the corresponding particle size range obtained from two monitoring stations (one at an urban background site, one at street level) for the days of the corresponding indoor measurements, air exchange rates estimated from occupantgenerated CO2 and an estimated penetration factor and deposition rate taken from the literature (see Supporting Information for details). From the average indoor background PN concentration, the background residential daily integrated exposure was calculated for the periods during which a home was occupied. A source event was defined as a sharp excursion in the indoor PN concentration. Events were analyzed if the indoor PN concentration within several minutes increased to a level at least twice as high as the indoor background PN concentration for at least 10 min. The time from the beginning of a rapid increase in the PN concentration until the concentration reached its peak was accepted as the duration of the source event (“source duration”). The duration of exposure to particles generated by a particular source event (“exposure duration”) was determined from the time when the PN concentration began to increase (same as for source duration), until it returned to a level that could be considered background concentration. The end of the elevated concentrations associated with an identified peak was often difficult to determine, since the indoor background concentration may have shifted while the source was active. Since these final concentrations are near back10242



Article

Figure 1. Two illustrative time-series of indoor PN concentrations and average particle diameters from two sites. The dashed lines denote the instrument’s lower particle size limit (10 nm diameter).

that matched very rare entries in the activity log books, for example the use of a fireplace, were categorized as special events. Recorded activities that could not be matched to altered particle concentrations were not analyzed further.

ground, their influence on the total contribution of the peak to the integrated exposure is small. To characterize each event’s net contribution to the total exposure, we calculated the integrated exposure to UFP generated by each event from the corresponding net indoor PN concentration. This was obtained as the difference between the actual indoor PN concentration and a unique event-specific indoor background PN concentration, determined as the average of the stable PN concentration at the beginning of an event and at the end of the elevated PN concentrations. Comparison of the average background concentrations estimated by different approaches and the event-specific background levels, as well as the treatment of peaks attributable to co-occurring events or to two events that closely followed each other, can be found in the Supporting Information. Source events were identified by associating the peaks with the activities recorded by the occupants in the activity log book. Activities were grouped into one of the following: cooking, toasting, candle burning, window opening, unknown events, and special events. Window opening was considered a source when it occurred for the purpose of occasional airing over a relatively short time without the co-occurrence of other indoor sources and resulted in a clear increase in the PN concentration. Peaks that could not be matched to an activity in the log books were categorized as unknown events. Peaks

■

RESULTS AND DISCUSSION Particle Concentrations and Average Particle Size. Table 1 shows the average PN concentrations and particle diameters calculated for the total measurement period, for periods when the homes were occupied, when the occupants were awake, or presumably asleep, as well as when the homes were unoccupied. Average PN concentrations obtained from the 56 homes were log-normally distributed, while average particle diameters were normally distributed. The homes were occupied on average for 19 h per day. Significantly elevated PN levels were observed when occupants were present and awake compared to all other occupancy periods. Lower PN concentrations were observed during the nights and during periods of vacancy. These results are in line with those of an extensive study in Ontario, Canada by Kearney et al.,27 who found that indoor UFP levels were lowest overnight, indicating the absence or reduction of indoor sources, while peaks around noon and 6 pm probably reflected cooking events. In the present study, slightly higher geometric mean and median PN concentrations were obtained when the 10243



Article

Table 2. Summary of the Occurrence and Duration of the Analyzed Source Events activity type cooking toasting candle burning special events unknown window opening

no. of events analyzed

no. of homes with the event

average no. of events per day (SD)a

average daily source duration (SD) (min/ d)a

average source duration per event (SD) (min)b

average daily exposure duration (SD) (h/d)a

average exposure duration per event (SD) (h)b

60 14 39

37 12 28

0.89 (0.4) 0.61 (0.2) 0.73 (0.38)

35 (22) 15 (11) 142 (108)

40 (24) 26 (18) 201 (108)

4.5 (2.7) 2.1 (1.0) 5.7 (2.9)

5.2 (2.4) 3.6 (2.1) 8.2 (3.5)

4

3

0.69 (0.31)

51 (44)

96 (90)

3.4 (1.9)

6.2 (4.5)

118 14

52 11

1.21 (0.63) 0.66 (0.25)

50 (62) 30 (27)

37 (36) 44 (36)

3.8 (3.2) 1.5 (1.2)

3.0 (2.1) 2.3 (2.0)

a Based on the number of homes where the particular type of event occurred (see third column). bBased on all events analyzed within a given type (see second column).

windows (each a separate group). A large number of “unknown” events (separate group) were observed (n = 118). These may reflect events that the occupants did not record or the penetration of particles from activities occurring in neighboring apartments. Although unlikely, a few of the unknown events of smaller magnitude may have been caused by changes in the outdoor UFP levels. However, data from outdoor monitoring stations, observations from the questionnaires, and comparison of the times of occurrence and the PN patterns of unknown events to those of identified events suggest that the majority of the unknown events were related to source events within the measured homes (see Supporting Information for detailed discussion on the potential sources of unknown events). Ironing, the use of a fireplace, and occasional smoking were each reported in one home. These clearly identifiable peaks were included in the group of “special events.” Reported activities that did not result in changes in the PN concentrations were boiling, microwave use, vacuuming, sweeping, air freshener in use, washing machine operation, tumbler drier operation, printer use, and making the bed. Thus, activities most strongly affecting UFP concentrations indoors were those related to cooking and combustion, less so to cleaning or use of electric appliances.20 However, the inability of the instruments to detect particles smaller than 10 nm may have underestimated the true UFP levels from the identified sources and may have missed elevated concentrations from some of the other known indoor sources. Elevated PN concentrations due to events lasted significantly longer than the corresponding source duration. Thus, the exposure to particles generated during an event lasted several hours after the event ceased. Kearney et al.27 indicated that an indoor source may cause lingering elevated PN levels for 5−10 h. The period of increasing PN concentrations during an event may not accurately match the source duration. The PN concentration for some particle sizes may begin to decrease even while the source is still producing particles. This occurs when the buildup of larger particles through coagulation results in loss rates for the smaller particles greater than the gain due to generation from the source.37 Integrated Exposure. The geometric mean daily integrated UFP exposure in the 56 residences was 334 × 103 cm−3 h/d (geometric standard deviation, GSD: 3.3; arithmetic mean: 667 × 103 cm−3 h/d; standard deviation, SD: 963 × 103 cm−3 h/d). On average, 88% (SD: 7.5%) of the total residential daily integrated exposure occurred while the occupants were awake (Figure 2). This is consistent with studies conducted in

occupants were away from home compared to when they were asleep. This may reflect higher outdoor concentrations of UFP during the day compared to the night (Figure S4).35,36 Occasionally higher PN concentrations during the night were observed. This may be caused by (1) residual particles from an earlier source event such as a burning candle,27 (2) an active source present during the night (two night events were recordedcandle burning throughout most of one particular night and a short unknown event), or (3) lack of information on the exact periods during which all occupants were asleep. All differences between PN concentrations during the various occupancy periods in Table 1 were statistically significant (p < 0.05; Wilcoxon signed-rank test; STATA software, release 11.2 for Windows, StataCorp LP, College Station, Texas, USA). The Supporting Information contains a discussion addressing the ratios of the average indoor PN concentration to the average indoor background concentration. The mean particle diameters during the various periods ranged from 70 to 90 nm: 70 nm when occupants were awake, 90 nm when they were asleep, and 73 nm during unoccupied periods (the differences between the three periods, except between awake and unoccupied periods, were statistically significant; p < 0.01; paired t test). During source events we consistently observed rapidly increasing PN concentrations together with decreasing average particle diameter (typically below 50 nm and often down to 20 nm) (Figure 1). Freshly generated combustion related particles are smaller than particles that have been present in the air for some time.26,37 Once a source event ceased, it was followed by a decrease in the PN level and a gradual increase in the average particle diameter. Decreasing PN levels may be attributable to diffusional deposition, ventilation, and coagulation. Coagulation alters the size distribution of particles as they age and is especially pronounced for smaller UFP and high PN concentrations.37,38 Larger average particle diameters were measured during the night, consistent with the absence of indoor sources and less traffic compared to daytime, leading to the presence of, on average, more aged aerosol (resulting from more time for particle growth due to condensation and coagulation).13 This increase in particle size during the night is amplified by the fact that particles in the 100−300 nm range have lower deposition rates and higher infiltration than particles

Ultrafine Particles: Exposure and Source Apportionment in 56 Danish

Recommend Documents