Full text of DOI:10.1080/10473289.2000.10464154

This article was downloaded by: [University of Connecticut]
On: 07 October 2014, At: 09:53
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer
House, 37-41 Mortimer Street, London W1T 3JH, UK
Journal of the Air & Waste Management Association
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/uawm20
Characterization of Indoor Particle Sources Using
Continuous Mass and Size Monitors
Christopher M. Long a , Helen H. Suh a & Petros Koutrakis a
a Harvard School of Public Health, Department of Environmental Health , Boston ,
Massachusetts , USA
Published online: 27 Dec 2011.
To cite this article: Christopher M. Long , Helen H. Suh & Petros Koutrakis (2000) Characterization of Indoor Particle
Sources Using Continuous Mass and Size Monitors, Journal of the Air & Waste Management Association, 50:7, 1236-1250,
DOI: 10.1080/10473289.2000.10464154
To link to this article: http://dx.doi.org/10.1080/10473289.2000.10464154
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose
of the Content. Any opinions and views expressed in this publication are the opinions and views of the
authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should
not be relied upon and should be independently verified with primary sources of information. Taylor
and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses,
damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection
with, in relation to or arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions

Long, Suh, and Koutrakis
TECHNICAL PAPER
ISSN 1047-3289 J. Air & Waste Manage. Assoc. 50:1236-1250
Copyright 2000 Air & Waste Management Association
Characterization of Indoor Particle Sources Using Continuous
Mass and Size Monitors
Christopher M. Long, Helen H. Suh, and Petros Koutrakis
Harvard School of Public Health, Department of Environmental Health, Boston, Massachusetts
ABSTRACT
sources of ultrafine particles characterized in this study
are indoor ozone/terpene reactions. Furthermore, EC/OC
data suggest that organic carbon is a major constituent of
particles emitted during indoor source events. Whether
exposures to indoor-generated particles, particularly from
large short-term peak events, may be associated with ad-
verse health effects will become clearer when biological
mechanisms are better known.
A comprehensive indoor particle characterization study
was conducted in nine Boston-area homes in 1998 in or-
der to characterize sources of PM in indoor environments.
State-of-the-art sampling methodologies were used to
obtain continuous PM2.5 concentration and size distribu-
tion particulate data for both indoor and outdoor air. Study
homes, five of which were sampled during two seasons,
were monitored over week-long periods. Among other data
collected during the extensive monitoring efforts were 24-
hr elemental/organic carbon (EC/OC) particulate data as
well as semi-continuous air exchange rates and time-
activity information.
This rich data set shows that indoor particle events
tend to be brief, intermittent, and highly variable, thus
requiring the use of continuous instrumentation for their
characterization. In addition to dramatically increasing
indoor PM2.5 concentrations, these data demonstrate that
indoor particle events can significantly alter the size dis-
tribution and composition of indoor particles. Source
event data demonstrate that the impacts of indoor activi-
INTRODUCTION
Numerous epidemiologic studies have shown a signifi-
cant association between fluctuations in ambient PM lev-
els and changes in daily mortality and morbidity.1-5
Because people spend approximately 85% of their time
indoors,6 it is widely recognized that a significant portion
of total personal exposure to ambient PM occurs in in-
door environments. Due to the presence of indoor par-
ticle sources, indoor exposures can also include particles
of indoor origin. Among the indoor particle sources iden-
tified in recent studies are cooking activities such as fry-
ing, sautéing, toasting, and baking;7-13 household
appliances such as gas-fired ranges and ovens, kerosene
heaters, wood-stoves, fireplaces, and humidifiers;12,14-16 and
general activities such as walking and even flailing one’s
arms.7,13,17 As a result of these indoor sources, indoor par-
ticle concentrations are often higher than ambient con-
centrations.18 It remains a possibility that exposures to
indoor particles (of ambient or indoor origin) may be as-
sociated with adverse health effects.
Despite the public health implications of indoor par-
ticle exposures, few studies have characterized sources of
indoor particles. Much of the information on indoor par-
ticles has been obtained from studies using traditional,
time-integrated sampling techniques. However, the tem-
poral variability of indoor particle sources demands the
collection of short-term particulate mass concentration
and speciation data, since indoor particle levels, size, and
composition may vary with source type. Indoor particles
ties are especially pronounced in the ultrafine (da
0.1 µm) and coarse (2.5 ≤ da ≤ 10 µm) modes. Among the
≤
IMPLICATIONS
This paper discusses the results of a comprehensive in-
door particle characterization study of nine Boston-area
homes. Data presented in this paper show that indoor
particle events can greatly elevate short-term indoor ex-
posures to PM2.5, particularly within tight, poorly venti-
lated residences. The impacts of indoor particle events
are especially pronounced in the ultrafine and coarse-
mode regions. In addition, this study provides evidence
that indoor chemical reactions are a potentially important
source of ultrafine particles under typical household con-
ditions. The health significance of these results will be-
come clearer when toxicological studies have elucidated
the biological mechanism for particulate health effects.
1236 Journal of the Air & Waste Management Association
Volume 50 July 2000

Long, Suh, and Koutrakis
are further affected by dynamic processes such as deposi-
tion, penetration, chemical reactions, and coagulation.
Only recently have studies begun to examine the effect
of these processes on indoor particle concentrations and
size distributions.
concentrations and their sources, and (3) investigate the
impact of indoor particle sources on the relationship be-
tween indoor and outdoor concentrations. A forthcoming
paper will use the time- and size-resolved particulate data to
address the issue of ambient particle infiltration.
Among the first was a pilot study conducted in four
homes in the Boston metropolitan area.7,19 Using continu-
ous indoor and outdoor volumetric size distribution data
and detailed time-activity information, Abt et al. 7,19 iden-
tified cooking, cleaning, and the movement of people as
the most important indoor particle sources in their study
homes. Using a simple physical-statistical model based
on indoor and outdoor particle volume concentrations,
the contributions of four indoor-activity classes––cook-
ing, cleaning, washing, and indoor work (e.g., walking,
field sampling, and other general activities associated with
physical movement)––were estimated for particles of spe-
cific sizes.19 Outdoor particles were found to be the pre-
dominant source of indoor fine (da ≤ 2.5 µm) particles
with cooking activities, such as oven baking and toast-
ing, being the only other significant sources of fine par-
ticles. In contrast, coarse particles had several indoor
sources, such as cooking, cleaning, and indoor work,
which were all shown to have significant effects on coarse
particle volume concentrations. This study, however, left
several issues unaddressed, since its small sample size did
not allow the relative importance of ambient penetration
and indoor deposition on indoor particle levels to be ex-
amined. In addition, the lack of continuous mass moni-
toring data prevented the impact of indoor particle sources
on mass concentration from being investigated.
METHODS
Study Design
Nine non-smoking Boston-area homes were chosen for
participation in the study. Study homes were primarily se-
lected based upon the willingness of homeowners to par-
ticipate in the study, given its intrusive nature and
demanding participation requirements. All homes were
located within 30 mi of downtown Boston in suburban
neighborhoods. Several of these homes were located within
a mile of heavily traveled roadways. From two to five people
lived in each home. Five homes had gas stoves, three homes
had electric stoves, and one home had both a gas-fired range
and an electric oven. Five of the nine homes were heated
with oil, while the remaining four had natural gas heating
systems. There was a similar split between study homes
with radiative heat and those with forced-air heat. Only
one of the nine study homes had a central air-condition-
ing and heating system (Home FOX1).
Study homes were primarily sampled during two time
periods—spring/summer 1998 and fall/winter 1998. Five
of the nine study homes were sampled during both of
these time periods. All homes were sampled a minimum
of six consecutive days on each sampling occasion, with
most homes sampled for at least 7 days and several for
longer periods. Table 1 lists the locations, sampling dates,
and sampling duration for each study home.
To address these issues, a more comprehensive follow-
up study was conducted in 1998. As part of this study, con-
tinuous indoor and outdoor size distribution and PM2.5 mass
concentration data were collected in nine homes in the met-
ropolitan Boston area. These data
Sampling equipment was placed in a central room in
the main living area of the study home (e.g., living room
or dining room) to facilitate air exchange with other rooms
Table 1. Sampling locations and dates.
were supplemented with time-inte-
grated fine particle speciation mea-
surements, including elemental/
organic carbon (EC/OC) concentra-
tions. This study also included
Home ID
Home Location
Season
Sampling Dates
Sampling
Start
Stop
Duration (days)
MAN1
Manchester-by-the-Sea
Newton
Winter
Spring
Fall
Spring
Winter
Summer
Summer
Winter
Summer
Fall
2/13/98
3/26/98
10/14/98
4/28/98
12/1/98
5/28/98
6/9/98
11/22/98
6/20/98
10/23/98
7/7/98
2/20/98
4/4/98
10/21/98
5/9/98
12/8/98
6/5/98
6/18/98
11/29/98
6/26/98
10/30/98
7/16/98
12/17/98
11/12/98
11/20/98
7
9
7
12
7
8
9
7
6
7
scripted indoor activities to assess the NEW1
individual contributions of specific
indoor activities to indoor particle
concentrations. In addition, semi-
WEL1
Wellesley
SWP1
BOX1
Swampscott
Boxford
continuous air exchange rates and
time-activity information were col-
NEW2
FOX1
Newton
lected for each home throughout
each sampling period.
The objectives of this paper are
to (1) describe the study design, (2)
present physical and chemical data
characterizing indoor particle
Foxboro
Summer
Winter
Winter
Winter
9
7
7
7
12/10/98
11/5/98
11/13/98
WEL2
SWP2
Wellesley
Swampscott
Volume 50 July 2000
Journal of the Air & Waste Management Association 1237

Long, Suh, and Koutrakis
in the main living area. Outdoor sampling equipment was
typically set up in the backyard or on the back porch of
the study homes. Homes were visited every 12 hr for the
duration of the sampling period. The purpose of these
visits was to (1) monitor equipment performance and
conduct flow checks, (2) change time-integrated samplers,
(3) review time-activity diaries with homeowners, (4)
download data from continuous sampling instruments,
and (5) perform scripted particle-generating activities
when homeowners were not home or when no particle-
generating activities were occurring.
Particle Size. Real-time size distribution measurements
were made using two particle-sizing instruments: the Scan-
ning Mobility Particle Sizer (SMPS) and the Aerodynamic
Particle Sizer (APS). As described elsewhere,7 these instru-
ments alternately sampled both indoor and outdoor air
from ports in a specially designed stainless steel sampling
manifold. The manifold consists of two identical sampling
arms, one that extends into the sampling room at breath-
ing height (approximately 5.5 ft) and another that ex-
tends through a window in the home to the outdoors.
Electronically controlled ball valves were used to rotate
between indoor and outdoor samples, and the manifold
sampling system was programmed to sample for three 5-
min intervals indoors followed by one 5-min interval
outdoors. The larger indoor sampling time was selected
based on the greater variability in indoor particle con-
centrations.7,8,12,13
The SMPS (Model 3934, TSI, Inc.) was used to con-
tinuously measure particle count concentrations in 46
discrete size bins between 0.02 and 0.5 µm. This instru-
ment consists of an electrostatic classifier (TSI Model 3071)
and a condensation particle counter ([CPC] TSI Model
3010). The APS (TSI Model 3310A) was used to continu-
ously measure particle count concentrations in 37 dis-
crete size bins between 0.7 and 10 µm.
Sampling Methods
Mass Concentration. Both continuous and time-integrated
sampling instruments were used to simultaneously mea-
sure indoor/outdoor (I/O) particle mass concentration
in each study home. The Tapered Element Oscillating
Microbalance (TEOM) (Model 1400A, Rupprecht &
Patashnick Co., Inc.) was used indoors and outdoors to
measure 5-min average PM2.5 concentrations.20 The TEOM
is a U.S. Environmental Protection Agency (EPA)-
approved continuous PM10 ambient particulate monitor,
and this instrument was adapted for PM2.5 sampling at 4
L/min (LPM) with the addition of a PM2.5 inlet/impactor.
Detection limits are approximately 5 µg/m3 for 10-min
average concentrations and 1.5 µg/m3 for 1-hr averag-
ing periods.21
Chemical Composition. Twenty-four hour indoor/outdoor
fine mass samples were collected for EC/OC analysis us-
ing filter pack samplers with a 2.5-µm cut size and a flow
rate of 10 LPM. These samplers were loaded with paired
(front and backup) pre-fired quartz fiber filters. However,
due to the uncertainty regarding the interpretation of data
from the backup filters,24 EC/OC concentrations reported
in this paper are solely for carbon measured on the front
filter. A parallel plate denuder containing carbon-impreg-
nated papers was used for the spring/summer 1998 study
homes to remove vapor-phase organic carbon prior to fil-
ter collection, but use of this denuder was discontinued
after extensive field testing indicated that the denuder
efficiency was significantly less than 100%.24 EC/OC
samples were stored at 0 °C until analysis by thermal/
optical reflectance (TOR)25 at the Desert Research Insti-
tute (DRI) (Reno, NV). Field blanks, which comprised ap-
proximately 10% of the total number of samples collected,
were used to estimate LODs of 0.27–0.58 µg/m3 for or-
ganic carbon and 0.04–0.14 µg/m3 for elemental carbon.
Twelve-hour time-integrated indoor/outdoor PM2.5
and PM10 samples were also collected using Harvard Im-
pactors (HIs).22,23 Samples were collected at a flow rate of
10 LPM on pre-weighed 37-mm Teflon filters (Teflo,
Gelman Sciences) for gravimetric analysis. An inlet
nozzle/impactor with a 50% cut-point of either 2.5 or
10 µm was used to fractionate particles by size. All filters
were on- and off-weighed twice using a Mettler MT-5
Microbalance (Mettler Toledo International, Inc.) in a
temperature- and relative humidity-controlled weighing
room after at least 48 hr of equilibration time. Filters
were weighed a third time if the two weights differed by
more than 5 µg. Field blanks comprised approximately
10% of the total number of collected samples. Blank cor-
rections were made when the blanks differed significantly
from 0 µg. Based on the variability of the field blanks
(e.g., 3 times the standard deviation of their mass in-
crease), the limits of detection (LODs) for PM2.5 and PM10
were determined to be less than 2 µg/m3. Replicate PM2.5
samples (n = 37) were collected in several homes in or-
der to assess sampler precision, and the coefficient of
variation (CV) was calculated as the standard deviation
of the difference between collocated samples divided by
the mean of the collocated samples and by the √2 to
adjust for the noise of the two samplers. The method
showed good precision with a CV of 4.6%.
Air Exchange Rates. Air exchange rates were measured in
each study home every 5 min using a constant sulfur
hexafluoride (SF6) source. This technique employs a
source that emits SF6 gas at a constant rate (6 mL/min)
and a monitor (Brüel & Kjær, Model 3425) that continu-
ously measures the SF6 concentration. The monitor,
1238 Journal of the Air & Waste Management Association
Volume 50 July 2000

Long, Suh, and Koutrakis
which detects the gas based on the principle of photo-
acoustic infrared spectroscopy, has a detection limit of 5
parts per billion (ppb).26 Air exchange rates were computed
using the 5-min SF6 concentration data, the known source
emission rate, and the house volume.7,27
used as the reference method because it is a widely used
and well-documented particulate sampling method.
R2 values of 0.93 and 0.92 were obtained for the re-
gressions of all matching 12-hr HI and TEOM indoor and
outdoor data, respectively, and they show the strong cor-
relation between these two methodologies for 12-hr PM2.5
data. However, as other recent studies have observed for
ambient measurements,28,29 the TEOM consistently mea-
sured lower indoor and outdoor particle concentrations
than the impactors did. This is reflected in the regression
slopes of 0.78 and 0.89 for the analyses of all indoor and
outdoor data, respectively. Although the TEOM exhibited
a low bias, this measurement bias was consistent and thus
allowed for the calculation of seasonal-specific correction
factors.
Housing and Time-Activity Information. Information on
housing characteristics was obtained using a detailed
housing questionnaire, which was previously developed
for use in the pilot study of four Boston-area homes.7
Time-activity information was recorded by the home oc-
cupants in 20-min intervals using a tabular, closed-form
time-activity diary, which was developed specifically for
this study. This diary consisted of a matrix of 20-min time
intervals and potential particle-generating activities. In
addition, space was provided for homeowners to write in
activities that were not already included in the compre-
hensive listing. These diaries were also used to record ven-
tilation conditions (i.e., opening and closing of windows
and doors, air conditioner usage, etc.).
Negative 12-hr PM2.5-10 concentrations were some-
times estimated from the difference between HI PM10 and
PM2.5 concentrations. These negative values primarily oc-
curred when PM10 and PM2.5 concentrations were very low
and near the LOD. They were not removed due to their
infrequent occurrence (which is supported by the posi-
tive 5th percentile values in Table 2). In addition, nega-
tive 1-hr PM2.5 concentrations measured by the TEOM were
also observed. These negative values occurred when the
instrument was physically jarred or following large in-
door cooking/combustion events when semi-volatile
material would slowly bake off the filter and cause a tem-
porary negative downswing. When identified, these data
were removed.
For the SMPS and APS size distribution data, com-
mercial software programs were used to convert raw count
data to volume or number concentrations (TSI SMPS
v. 2.3 and APS EXTRA v. 1.1).7 In addition, APS data were
corrected for depositional losses in the sampling mani-
fold based on a regression equation developed from pre-
vious laboratory tests.7,30 No corrections were made to the
SMPS data, which were found to have no significant mani-
fold losses in these laboratory tests.7,30
Scripted Activities. This study not only relied on the ev-
eryday activities of home occupants to provide particle
events for characterization but also included scripted in-
door activities. The typical household is characterized
by periods of little or no particle-generating activities
when homeowners are absent or inactive as well as ac-
tive periods when homeowners engage in multiple si-
multaneous activities, such as episodic bursts of cooking
and cleaning. The absence of particle-generating activi-
ties can leave few particles for study, especially when
ambient particle concentrations are very low, while the
simultaneous occurrence of many particle-generating
activities is also problematic with respect to the appor-
tionment of measured particles to specific activities. In
order to increase the number of distinct indoor particle
events for characterization, scripted activities were per-
formed in study homes during periods when there were
no other identifiable particle sources. Scripted activities
performed in study homes included cooking and clean-
ing activities, resuspension events, ventilation changes,
and household appliance usage.
Size distribution data are typically reported in this pa-
per as particle volume (PV) concentration data in units of
µm3/cm3 for four particle size ranges: 0.02–0.1 (PV0.02–0.1
)
and 0.1–0.5 µm (PV0.1–0.5) for SMPS data, and 0.7–2.5
(PV0.7–2.5) and 2.5–10 µm (PV2.5–10) for APS data. These size
ranges were chosen so that the large data set could be
summarized and observations could be specifically made
for ultrafine particles (PV0.02–0.1), accumulation-mode par-
ticles (PV0.1–0.5), and coarse-mode particles (PV2.5–10). No data
are reported for the 0.5–0.7-µm size interval, since previ-
ous studies have demonstrated that neither the SMPS nor
the APS accurately measures concentrations of particles
of these sizes.30-32 As previously mentioned, SMPS and APS
data were measured as 5-min average concentrations
where indoor measurements were made at the 0-, 5-,
Data Validation and Analysis
Initially, all data were checked for quality assurance. The
availability of replicate data from independent method-
ologies allowed the performance of several instruments
to be evaluated and correction factors to be determined.
For example, because the TEOM is known to lose semi-
volatile material due to the heating of the sample filter,28,29
its performance for both indoor and outdoor aerosols was
evaluated by regressing its 12-hr average PM2.5 concentra-
tion data on the corresponding 12-hr HI data. The HI was
Volume 50 July 2000
Journal of the Air & Waste Management Association 1239

Long, Suh, and Koutrakis
10-, 20-, 25-, 30-, 40-, 45-, and 50-min intervals of each
hour, while outdoor measurements were made at the 15-,
35-, and 55-min intervals. Hourly averaged indoor con-
centrations are thus computed from 45 min of data, while
outdoor concentrations are calculated based on 15 min of
data.
available as 1-hr and 12-hr average PM2.5 concentrations
and 12-hr PM10 and PM2.5-10 concentrations. The mean
indoor hourly PM2.5 concentration was 11.8 µg/m3 (SD =
20.7), while the mean 12-hr indoor HI PM2.5 concentration
was 11.9 µg/m3 (SD = 9.6). The short-term impacts of in-
door sources are reflected both in the larger standard de-
viation for the hourly PM2.5 data in comparison to the 12-hr
measurements and in the greater maximum indoor PM2.5
concentration for the hourly data (471.9 vs 85.5 µg/m3 for
the 12-hr HI samples). The mean outdoor hourly PM2.5
concentration was 11.1 µg/m3 (SD = 9.5), while the mean
outdoor 12-hr PM2.5 concentration was 11.1 µg/m3 (SD =
6.8). The larger indoor standard deviations reflect the greater
variability of indoor concentrations, which can be attrib-
uted to the episodic contributions of indoor sources.
Indoor PV0.02–0.1, PV0.1–0.5, PV0.7–2.5, and PV2.5–10 mean
hourly concentrations were 0.71 µm3/cm3 (SD = 1.8), 6.3
µm3/cm3 (SD = 9.8), 2.3 µm3/cm3 (SD = 9.2), and 2.4
The Statistical Analysis System (SAS Institute, Cary,
NC) was used for data analysis. Among the statistics em-
ployed were descriptive statistics including box plots and
histograms, Spearman correlation coefficients, and analy-
sis of variance (ANOVA).
RESULTS
Summary of Indoor and Outdoor
Particulate Concentration Data
Table 2 provides summary statistics for the indoor and
outdoor particulate concentration data collected during
this study. For most homes, mass concentration data are
Table 2. Summary of particulate concentration data for all study homes.a
Sample Type
N
Mean SDb
Minimumc
5th
Median
95th
Maximum
Percentile
Percentile
Integrated PM
(12-hr samples)
Indoor PM2.5
211
212
211
210
107
107
11.9 9.6
19.4 12.7
7.6 6.2
11.1 6.8
12.7 7.5
4.6 3.7
2.1
2.8
-4.7
0.47
3.2
3.3
3.1
0.10
3.2
4.7
0.91
9.0
16.0
6.1
8.3
9.8
32.1
42.2
21.4
27.4
29.0
10.8
85.5
95.3
46.3
46.4
32.6
21.2
Indoor PM10
Indoor PM2.5-10
d
Outdoor PM2.5
e
Outdoor PM10
d,e
Outdoor PM2.5-10
-3.8
3.8
Continuous PM
(1-hr averages)
f,g
Indoor PM2.5
2340
2075
11.8 20.7
11.1 9.5
-54.0
-2.6
0.51
1.8
7.5
8.5
32.8
29.6
471.9
221.3
f,g
Outdoor PM2.5
Particle Volume: Continuous Size Distribution
(1-hr averages)
Indoor PV(0.02-0.1)
2567
2567
2377
2377
2565
2565
2375
2375
0.71 1.8
6.3 9.8
2.3 9.2
2.4 3.8
0.49 0.59
6.3 5.1
2.5 2.7
2.7 4.7
0.024
0.27
0.069
0.041
0.029
0.23
0.073
0.84
0.19
0.14
0.071
1.03
0.29
0.44
0.32
3.7
1.2
1.4
0.37
4.6
1.9
19.2
6.9
33.4
215.3
374.4
69.3
13.6
64.5
Indoor PV(0.1-0.5)
Indoor PV(0.7-2.5)
Indoor PV(2.5-10)
h
h
7.4
Outdoor PV(0.02-0.1)
Outdoor PV(0.1-0.5)
Outdoor PV(0.7-2.5)
1.29
17.7
8.3
h
0.13
0.098
1.4
1.8
28.2
174.2
h
Outdoor PV(2.5-10)
7.9
EC/OC
(24-hr samples)
Indoor Total Carbon
Indoor OC
Indoor EC
Outdoor Total Carbon
105
105
105
103
103
103
8.5 3.1
7.7 2.9
0.85 0.41
4.0 1.6
3.1 1.2
0.97 0.51
3.1
2.7
0
0.47
0.44
0.053
4.7
4.3
0.20
1.2
0.98
0.23
8.1
7.2
0.76
3.9
2.9
0.85
14.5
12.4
1.8
7.2
5.1
31.1
29.5
3.0
8.7
7.1
Outdoor OC
Outdoor EC
2.1
2.8
Notes: aHIs, TEOM, and EC/OC data are in µg/m3, while SMPS/APS data are in µm3/cm3. bSD refers to pooled standard deviation. cNegative values are an artifact of the collection and
estimation methods (see text for discussion). dEstimated using the difference of PM10 and PM2.5.eOutdoor PM10 HIs only collected during fall and winter sampling. fTEOM data have
g
been corrected using results of seasonal-specific regressions with HI data. For logistical reasons, TEOM measurements were not made indoors at home NEW1 (spring sampling
only) and outdoors at homes MAN1 and WEL1 (spring sampling only). hAll but 1 of 9 days of APS data from home NEW1 (spring sampling only) were lost due to equipment failure.
1240 Journal of the Air & Waste Management Association
Volume 50 July 2000

Long, Suh, and Koutrakis
µm3/cm3 (SD = 3.8), respectively. Outdoor PV0.02–0.1
,
Summary of Air Exchange Rate Data
PV0.1–0.5, PV0.7–2.5, and PV2.5–10 mean hourly concentrations
were 0.49 µm3/cm3 (SD = 0.59), 6.3 µm3/cm3 (SD = 5.1),
2.5 µm3/cm3 (SD = 2.7), and 2.7 µm3/cm3 (SD = 4.7), re-
spectively. With the exception of the PV2.5–10 data, the stan-
dard deviations for the indoor data were larger than those
for the outdoor data. The higher variability in indoor con-
centrations can again be explained by the short-term im-
pacts of indoor source events. The outdoor PV2.5–10 mean
hourly concentration and its standard deviation were
greater than their corresponding indoor values, which is
largely due to the impact of several large outdoor coarse-
particle events including grass-cutting and leaf-blowing.
Twenty-four hour EC/OC samples provide informa-
tion on the composition of indoor particles. As shown in
Table 2, the mean indoor organic carbon concentration
was 7.7 µg/m3 (SD = 2.9), while the mean indoor elemen-
tal carbon concentration was 0.85 µg/m3 (SD = 0.41).
Outdoor organic carbon concentrations were significantly
lower (p < 0.0001) with a mean value of 3.1 µg/m3 (SD =
1.2), while elemental carbon concentrations were on av-
erage slightly higher, with a mean value of 0.97 µg/m3
(SD = 0.51). On average, organic carbon thus comprised
approximately 90% and 76% of the total carbon measured
indoors and outdoors, respectively.
For all data, hourly air exchange rates ranged between
0.11 and 20.40 hr-1 with mean and median values of 1.22
and 0.74 hr-1, respectively. Air exchange rates varied sig-
nificantly (p < 0.0001) with season, with the highest mean
air exchange rates occurring in the summer sampling pe-
riod (see Figure 1). In addition, there was variability within
seasons due to differing home-to-home ventilation con-
ditions. For example, homes SWP1 (mean = 3.17 hr-1) and
NEW2 (mean = 4.74 hr-1) were extremely well-ventilated
during the summer, as windows and doors were left open
for prolonged periods of time, including overnight. As a
result, air exchange rates in these well-ventilated homes
were higher than those in the summer in homes BOX1
(mean = 0.79 hr-1) and FOX1 (the only home with a cen-
tral air-conditioning system; mean = 0.20 hr-1), which
despite the warm temperatures, had their windows closed
for most or all of the monitoring period. Winter air ex-
change rates also varied by home. Air exchange rates
greater than 1 exchange/hr were common in home SWP2
(mean = 1.45 hr-1), where a bedroom window and doors
were left open during the daytime, and in home BOX1
(mean = 1.89 hr-1), which is more than 300 years old. Large
indoor/outdoor temperature differences likely caused air
exchange rates to be significantly higher (p < 0.0001)
Figure 1. Box plots of hourly air exchange rates stratified by season.
Volume 50 July 2000
Journal of the Air & Waste Management Association 1241

Long, Suh, and Koutrakis
during the winter than during the summer for home
BOX1. In contrast, winter air exchange rates were consis-
tently less than one for the other study homes.
activities and that were elevated above background par-
ticle levels. For these indoor source analyses, all concen-
tration data were corrected for background concentrations.
On average, the five largest peak PM2.5 concentrations were
observed during cooking activities—gas-fired oven bak-
ing (101 185 µg/m3), sautéing (66 95 µg/m3), toasting
(55 120 µg/m3), frying (41 43 µg/m3), and stir-frying
(37 31 µg/m3). The most important indoor sources of
coarse-mode peak events were found to be dusting (82
70 µm3/cm3), vigorous walking (29 19 µm3/cm3), and
sautéing (23 23 µm3/cm3). The large standard deviations
in the particle source contributions indicate that the
strength of indoor particle events is highly variable. This
may be explained by several factors including (1) non-
constant source conditions (e.g., cooking parameters such
as oven temperature and cleanliness, food type, and cook-
ing time33,34); (2) varying home conditions including air
exchange rates, kitchen fan usage, and dust loadings;17,35
and (3) the proximity of the monitoring equipment to
the source.36
Characterization of Indoor Particle Sources
Continuous particle concentration data and time-activ-
ity diaries were used to identify indoor sources of par-
ticles. Figure 2 shows PM2.5 mass concentrations measured
during two days in July 1998 in home FOX1. These data
demonstrate that indoor particle events tend to be brief
and intermittent, which leads to highly variable indoor
particulate concentrations. In this plot, the indoor PM2.5
concentration is characterized by large, short-term (e.g.,
minutes to hours) peaks and subsequent decay periods
from episodic indoor particle events superimposed on a
more slowly changing (e.g., hours to days) background
concentration that is likely comprised of particles of am-
bient origin. Similar results were found for other homes.
Figure 2 and data from the other study homes sup-
port previous findings that cooking, cleaning, and many
general activities are important sources of particles in in-
door residential environments.7,18 Table 3 summarizes the
maximum PM2.5 mass and size-specific volume concen-
trations for the indoor source types identified in this study
using data from more than 200 indoor particle events.
Indoor particle events were defined to include only those
data that could be attributed to specific individual source
Our findings substantiate the observations of Abt et
al.7 that indoor sources predominantly emit particles in
the ultrafine and coarse-mode regions (see Figure 3). This
figure presents bar charts depicting the average volumet-
ric size distributions for several of the indoor activities
reported in Table 3. In this figure, PV10 is the sum of the
SMPS (PV0.02–0.5) and APS (PV0.7–10) volume concentration
Figure 2. Continuous indoor and outdoor PM2.5 data (20-min moving average) from Home FOX1 (July 9–10, 1998).
1242 Journal of the Air & Waste Management Association
Volume 50 July 2000

Long, Suh, and Koutrakis
Table 3. Characteristic size distributions and maximum concentrations of indoor particle sources.a,b
Size Statistics
Maximum Concentrationsc,d
Particle Source
N
Indoor Activity Backgrounde
PM2.5
PV0.02-0.1
Mean SD
PV0.1-0.5
PV0.7-2.5
Mean SD
PV2.5-10
Mean
Mean VMD
Mean VMD
Mean
SD
Mean
SD
SD
(µm)
(µm)
Cooking
Baking (Electric)
Baking (Gas)
Toasting
8
24
23
4
0.189f
0.107f
0.221f
0.224f
14.8
101.2
54.9
29.3
65.6
37.2
40.5
14.8
7.4
2.2
18.2
5.0
4.5
3.2
5.8
3.6
2.3
2.0
19.0
6.6
2.8
5.5
4.9
2.9
2.7
12.7
45.7
30.9
16.0
16.2
12.6
15.1
5.6
14.9
87.5
74.8
30.3
31.0
16.3
18.5
1.8
1.1
3.1
1.3
10.3
9.6
1.5
5.0
1.1
1.6
23.2
8.2
2.8
3.6
1.2
9.9
184.9
119.7
43.4
95.4
31.4
43.2
5.2
0.138f
0.222f
4.0
1.1
Broiling
0.114f
0.236f
0.60
7.4
0.57
6.2
1.1
Sautéing
13
3
0.184f, 3.48g
0.135f
0.223f, 2.93g
0.277f
22.9
12.0
3.3
Stir-frying
3.8
5.5
Frying
20
2
0.173f
0.159f
0.223f
0.205f
2.2
2.4
Barbecuing
Cleaning
0.97
1.4
5.0
Dusting
11
10
5
5.38g
3.86g
0.097f
3.53g
2.79g
0.238f
22.6
6.5
22.6
3.9
0.01
0.06
5.0
0.01
0.08
8.2
0.14
0.43
6.0
0.19
0.49
12.4
5.0
1.8
6.4
1.4
82.2
6.7
70.3
3.7
Vacuuming
Cleaning with Pine Sol
General Activities
11.0
10.2
0.32
0.16
1.2
1.4
Walking Vigorously (w/Carpet) 15
3.96g
4.25g
4.28g
0.311f
3.18g
2.63g
2.93g
0.224f
12.0
8.0
9.1
6.6
0.04
0.08
0.01
0.78
0.13
0.15
0.05
0.64
0.19
0.46
0.23
36.9
0.27
0.71
0.29
43.9
6.7
1.3
5.0
0.80
0.58
2.7
29.1
8.0
3.9
1.9
19.4
5.6
2.1
1.8
Sampling w/Carpet
Sampling w/o Carpet
Burning Candles
52
26
7
4.8
3.0
0.63
4.8
28.0
18.0
Notes: aAll concentration data corrected for background particle levels.bIncludes only individual particle events that were unique for a given time period and could be detected above background
particle levels. cPM concentrations in µg/m3; PV concentrations in µm3/cm3.dMaximum concentrations computed from 5-min data for each activity.eBackground data are for time periods imme-
diately prior to the indoor event. fSize statistics calculated for PV0.02-0.5 using SMPS data. gSize statistics calculated for PV0.7-10 using APS data.
Figure 3. Mean volumetric particle size distributions for typical indoor activities (all data corrected for background).
Volume 50 July 2000
Journal of the Air & Waste Management Association 1243

Long, Suh, and Koutrakis
data for particles less than 10 µm in diameter. This fig-
ure clearly shows that ultrafine particles represent a major
portion of particle emissions from typical cooking ac-
tivities, including, on average, more than 50% of the
particles generated during 24 gas-fired oven events and
four broiling events. Coarse particles were generated by
the physical stirring of foods in fats and liquids that oc-
curs during sautéing, with nearly 50% of the particles
generated during 13 sautéing events being from the
coarse mode (see Figure 3). Furthermore, Table 3 con-
trasts the volume median diameter (VMD) for each source
type with those for time periods immediately prior to
the event, and it further confirms that emissions from
indoor particle events can have significant impacts on
indoor size distributions. For example, cooking activi-
ties generated sufficient amounts of ultrafine particles
such that the VMD during these events was characteris-
tically lower than during background periods immedi-
ately prior to these events (0.11–0.19 vs 0.21–0.28 µm).
The VMD was substantially elevated by sautéing events
as well as by resuspension events such as vigorous walk-
ing, dusting, vacuuming, and sampling (3.5–5.4 µm vs
2.6–3.5 µm).
background, respectively. Figure 4 shows data from a
walking event in home WEL1, which resulted in a short-
term peak PM2.5 concentration of approximately 25 µg/
m3. PM2.5 concentrations subsequently decayed over a
period of an hour to background levels. Furthermore,
statistically (p = 0.02) higher concentrations of particles
were found for sampling activities conducted in a car-
peted sampling room vs those occasions when the sam-
pling room had a hardwood or linoleum floor (maximum
concentrations of 8.0 6.6 vs 4.8 3.0 µg/m3 for car-
peted and non-carpeted sampling rooms, respectively).
This finding supports the conclusion of previous studies
that carpets can act as long-term reservoirs for
suspendable particles.17,37
Both PM2.5 and particle sizing data from several study
homes also provide evidence of significant ultrafine par-
ticle generation from indoor ozone/terpene reactions, as
hypothesized by Weschler and Shields.38 The generation
of aerosol products through ozone/terpene reactions has
been well-documented in laboratory studies39,40 and more
recently characterized in forest atmospheres41,42 and simu-
lated indoor environments.38 Recent laboratory and field
studies41-43 have identified organic aerosol products to in-
clude aldehydes and organic acids. In this study of resi-
dential indoor environments, ultrafine particles were
generated in five of six cleaning events using Pine Sol, a
commercial cleanser whose primary ingredient is α-pinene
(see Table 4). Three of the events were scripted activities
where hardwood and linoleum floors were mopped with
These data show that indoor resuspension events
such as walking, cleaning, and dusting are potentially
important sources of PM2.5 as well as coarse-mode par-
ticles (see Table 3). For 11 dusting and 15 vigorous walk-
ing events, mean peak PM2.5 concentrations were
approximately 23 (SD = 23) and 12 (SD = 9.1) µg/m3 above
Time (EST)
Figure 4. A resuspension event from home WEL1.
1244 Journal of the Air & Waste Management Association
Volume 50 July 2000

Long, Suh, and Koutrakis
Table 4. Summary of indoor residential ozone/terpene events.a
Background
Terpene Events
Event Type
Home
Date
Activity Timeb Mean AER
(hr-1)
Ambient Ozone Mean Volume Mean Number Mean PM2.5 Max. Volume Max.Number Conc.d m Max PM2.5 Conc
Conc.c (ppb) Conc.d (µm3/cm3) Conc.d(#/cm3) Conc.d (µg/m3) (µm3/cm3)
(#/cm3)
(µg/m3)
Mopping with Pine Sole
Mopping with Pine Sol
FOX1
NEW2
WEL1
BOX1
BOX1
7/10/98
10/29/98
12/2/98
6/11/98
11/24/98
16:30–16:45
13:30–13:40
16:40–16:50
15:20–16:00
14:40–15:20
0.53 0.25
0.94 0.12
0.56 0.06
0.50 0.04
2.02 0.29
44–48
17–25
1–11
3.4
1.1
1.6
5.5
0.6
1826
1080
2135
10,428
600
6.0
3.6
37.2
4.2
2.6
7.5
1.4
189,881
63,759
21,272
82,103
12,571
38
13
7
Mopping with Pine Sole
Cleaning Toilet with Pine Sol
Cleaning Toilet with Pine Sol
-3.0
10.2
2.3
34–52
32–34
13
9
Notes: aNot including data from a scripted mopping event in home SWP2 (11/19/98; 9:20–9:30) where no particle generation was observed. bAs reported in time-activity diaries. cRange of adjacent hourly data from closest available
state/local ambient monitoring site, as obtained from EPA AIRS website (www.epa.gov/airs). dCalculated using 5-min SMPS (0.02–0.5 µm) data only. eWindows/doors were opened for approximately 1/2 hr prior to commencement of terpene
experiments in order to increase indoor ozone concentrations with enhanced air exchange.
a dilute Pine Sol solution. The other two events were clean-
ing activities performed by a homeowner who used Pine
Sol to disinfect a bathroom toilet.
A key finding of this study is that more than 50% of
the particles (by volume) generated during these indoor
events were ultrafine particles (see Figure 3). During these
events, number concentrations increased dramatically
(peak increases ranged from 7- to 100-fold; see Table 4).
These increases in particle number resulted in concomi-
tant increases in the PM2.5 mass concentration of approxi-
mately 3–32 µg/m3, revealing the strong impact of particle
number on mass concentration (see Table 4). The largest
event, with a peak PM2.5 concentration increase of 32
µg/m3 and a peak number concentration increase of ap-
proximately 188,000 particles/cm3, occurred during sum-
mer sampling (July) in home FOX1. During this event,
local ambient O3 concentrations were moderately elevated
(44–48 ppb; note that several windows were opened for
half an hour before mopping began) and the air exchange
rate was fairly low during cleaning (mean = 0.53 hr-1),
two conditions that probably contributed to the magni-
tude of the event (see Table 4).
Figures 5(a)–(c) illustrate the impact of ultrafine par-
ticle generation via ozone/terpene chemistry in home
FOX1, in which 10 min of mopping with Pine Sol influ-
enced indoor particle concentrations for more than 8 hr.
Figure 5(a) shows the rapid increase in ultrafine particle
numbers and the subsequent shift in the number distri-
bution with time, which are characteristic of these ozone/
terpene events. As shown in Figure 5(b), ultrafine particle
formation initially results in a dramatic decrease in the
Figure 5. Particle distribution profiles for the FOX1 Pine Sol event.
VMD, followed by a gradual increase in the VMD as the
result of condensation and coagulation processes. The
ultrafine volume concentration (PV0.02–0.1) peaked at ap-
proximately 20 µm3/cm3, at which point ultrafine par-
ticles comprised nearly 100% of the total indoor number
concentration and more than 70% of the total volume con-
centration (see Figure 5[c]). Furthermore, the similarity
between the volume concentration (PV0.1–0.5) and mass
concentration (PM2.5) data in Figure 5(b) suggests that the
density of these particles is close to 1, which is similar to
that of ambient organic PM44 and is thus consistent with
ozone/terpene particle generation.
Volume 50 July 2000
Journal of the Air & Waste Management Association 1245

Long, Suh, and Koutrakis
Sources of Indoor EC/OC
non-denuded samples. It is important to note that a factor
of 1.4 was used to convert organic carbon concentrations
to organic particle concentrations, and this is based on
analyses of ambient particles.6,45 Although a lower factor
might be more appropriate for indoor particles, which
mainly consist of oxygenated hydrocarbons rather than
aliphatic hydrocarbons,46-49 this would not substantially
change the finding that indoor carbon samples (and par-
ticularly the non-denuded samples) often had higher car-
bon levels than the total mass of the corresponding PM2.5
samples.
These differences between the denuded and non-
denuded samples and between the indoor and outdoor
samples have several potential implications: (1) for both
denuded and non-denuded carbon samples, the quartz
filters not only collected particulate organic carbon but
also adsorbed gaseous organic carbon (conversely, there
also exists the possibility that semi-volatile organic car-
bon was lost from the HI PM2.5 samples and this explains
the disparities between the two sets of data); (2) although
not 100% efficient, the denuder did appear to reduce
the gas-phase organic carbon concentration and its sub-
sequent adsorption onto the quartz fiber filters (field tests
of these denuders yielded an average collection efficiency
of approximately 70% with large diurnal variability);24
and (3) both gas-phase and particulate-phase organic
The 24-hr EC/OC data suggest that organic PM had sig-
nificant indoor sources in these study homes, while in-
door elemental carbon was primarily of ambient origin.
These ideas are illustrated by Figure 6, which presents I/O
ratios for all denuded and non-denuded EC/OC data. This
figure shows that I/O ratios for organic carbon are typi-
cally on the order of 2–4 (for both denuded and non-
denuded samples), while I/O ratios for elemental carbon
are frequently less than 1. However, some caution must be
taken in interpreting these organic carbon data due to mea-
surement uncertainties associated with the use of quartz
fiber filters to measure particulate organic carbon.24 Spe-
cifically, indoor total carbonaceous particle concentrations
(the sum of elemental carbon and organic particle concen-
trations) were greater than the corresponding 24-hr HI PM2.5
concentrations for 58 of 105 sampling days with valid data.
For outdoor data, this was only true for 6 of 105 sampling
days. As mentioned earlier, a carbon denuder was used
during the first half of sampling. The denuded indoor
samples were found to have higher concentrations of total
carbonaceous particles than PM2.5 had in 24% (14 of 58) of
samples compared to 94% (44 of 47) of non-denuded
samples. For the outdoor samples, only 1 of 58 denuded
samples had higher concentrations of total carbonaceous
particles than PM2.5 had, and this compares with 5 of 47
Sample Type
Figure 6. Box plots of elemental and organic carbon indoor/outdoor ratios (with and without denuder).
1246 Journal of the Air & Waste Management Association
Volume 50 July 2000

Long, Suh, and Koutrakis
carbon are emitted from indoor source types, although
method uncertainties make it difficult to quantify the
partitioning of organic carbon between the particulate
and gaseous phases. Whether it is associated with the
gaseous or particulate phase, it is apparent from Figure 6
that organic carbon is a major constituent of indoor
source emissions.
8:00 a.m. to 11:00 p.m.). Nighttime periods were defined
as those times when people were asleep and thus include
only non-source data (typically 1:00 a.m. to 6:00 a.m.).
As expected, the hourly I/O correlations for all four size
intervals were higher for the nighttime data than for the
daytime data (see Figure 7), suggesting that indoor sources
act to decrease I/O associations (although, due to the size
of the data set, all correlation coefficients were highly sig-
nificant with p < 0.0001). In a similar finding, the I/O
correlations were found to be 0.66 and 0.78 for daytime
and nighttime PM2.5 data, respectively. The largest differ-
ences between correlation coefficients were for the
ultrafine (0.55 vs 0.87) and coarse-mode regions (0.21 vs
0.65), further emphasizing the role of indoor activities as
significant sources of ultrafine and coarse-mode particles.
In addition, for both daytime and nighttime data, corre-
lation coefficients for the ultrafine and coarse-mode par-
ticles were lower than those for the two other size ranges,
and this is most likely due to the higher deposition rates
of these particles.
I/O Relationships
The relationship between indoor and outdoor particle
concentrations was found to be highly influenced by the
short-term impacts of indoor particle events. For all hourly
data (n > 2000), indoor concentrations were found to be
highly correlated with corresponding outdoor concentra-
tions. Indoor and outdoor PM2.5, PV0.02–0.1, PV0.1–0.5
,
PV0.7–2.5, and PV2.5–10 correlations were found to have highly
significant (p ≤ 0.0001) Spearman correlation coefficients
of 0.69, 0.63, 0.82, 0.69, and 0.35, respectively. The low-
est correlation coefficient for the coarse-mode fraction was
likely the result of both higher depositional losses as well
as larger contributions of indoor sources.7
I/O ratios also reflect the contributions of indoor
sources. For daytime periods, I/O ratios for PM2.5 data ranged
from 0.03 to 257 with a mean of 2.4 14. For nighttime
non-source periods, I/O ratios for PM2.5 data ranged from
0.03 to 3.7 with a mean of 0.74 0.41. Figure 8 shows
frequency distribution plots of hourly I/O ratios for the
When data were stratified into daytime and night-
time periods, the effects of indoor activities and deposi-
tion losses on I/O correlations could be better assessed.
Daytime periods were defined as time periods when
homeowners were awake and potentially active (typically
Size Interval (µm)
Figure 7. Nighttime vs daytime indoor/outdoor correlation coefficients.
Volume 50 July 2000
Journal of the Air & Waste Management Association 1247

Long, Suh, and Koutrakis
four size intervals where data have been stratified by both
daytime and nighttime periods and by the magnitude of
the I/O ratio. For each of the size intervals, daytime I/O
ratios frequently exceed 1 and often 2, while nighttime
I/O ratios exceed 1 only rarely. Furthermore, when day-
time data were stratified by air exchange rate in simple
ANOVA models, PV0.02–0.1 and PV2.5–10 concentrations were
shown to be significantly greater when air exchange rates
were less than 1 as compared to when they were greater
than 1 (p values of 0.0003 and 0.033, respectively). These
results suggest that the impacts of indoor sources are even
more pronounced at low air exchange rates when indoor
residence times are longer and provide sufficient time for
particles from indoor sources to accumulate.7 This find-
ing also provides further evidence that the impacts of in-
door sources are especially pronounced in the ultrafine
and coarse-mode regions (similar ANOVA models yielded
non-significant p values of 0.50 and 0.66 for PV0.1–0.5 and
PV0.7–2.5 data), respectively.
CONCLUSIONS
This study utilized a novel study design, including both
continuous PM2.5 mass concentration and size distribution
measurements as well as the performance of scripted in-
door activities, to provide data for the characterization of
indoor sources of particles. This is one of the first residen-
tial indoor air studies where both indoor and outdoor PM2.5
were measured continuously, and these measurements
show that indoor fine particle events tend to be brief, in-
termittent, and highly variable. PM2.5 peak (5-min aver-
age) concentrations as high as 472 µg/m3 were measured
in study homes. Although indoor activities such as cook-
ing and cleaning can elevate short-term indoor PM2.5 con-
centrations as much as several orders of magnitude above
background, even the most dominant indoor events can-
not be quantified using the more traditional time-integrated
(e.g., 12- or 24-hr) samples.
Due to the contributions of indoor sources, indoor
exposures can have significant impacts on short-term
Figure 8. Frequency distribution (%) of hourly indoor/outdoor ratios stratified by daytime (e.g., source) vs nighttime (e.g., non-source) periods.
1248 Journal of the Air & Waste Management Association
Volume 50 July 2000

Long, Suh, and Koutrakis
personal PM2.5 exposures. Since several recent studies have
demonstrated associations between short-term PM exposures
and acute health outcomes, the characterization of short-
term indoor particle concentrations is of key importance in
our effort to assess particle health effects.50-53 In addition,
indoor sources can potentially bias integrated personal PM2.5
exposures, particularly in homes that are poorly venti-
lated.54 Indoor particle events can potentially have the
strongest influence on short-term and integrated personal
PM2.5 exposure for residents of tight, insulated homes in
colder climates (such as the Northeast in winter) as well as
residents of homes with central HVAC systems across the
United States. The impact of intermittent indoor events
on total personal PM2.5 exposures would be expected to be
of lesser importance in well-ventilated homes, such as those
where windows and doors are routinely left open. How-
ever, despite including predominantly well-ventilated Cali-
fornia homes (geometric mean air exchange rate = 0.97
2.18 hr-1), the EPA Particle Total Exposure Assessment Meth-
odology (PTEAM) study results also suggest the important
contributions of indoor activities such as cooking, smok-
ing, and cleaning activities to personal PM exposures.9 Due
to the scarcity of toxicity data for indoor and outdoor par-
ticles, the public health implications of indoor PM expo-
sures are not currently known.
Emissions from indoor sources can also significantly
alter the size distribution and composition of indoor par-
ticles. More than 200 distinct indoor particle events were
identified so that indoor source contributions could be
quantified and unambiguously attributed to individual
indoor activities. These data demonstrate that the im-
pact of indoor activities is especially pronounced in the
ultrafine and coarse-mode regions, such that indoor-
generated particles in these size ranges can dominate in-
door concentrations. Furthermore, 24-hr EC/OC samples
collected during this study suggest that particulate or-
ganic carbon is an important constituent of indoor par-
ticulate emissions.
30 µg/m3 during one of these events. PM2.5 concentrations
were also elevated by resuspension events, which were
found to be important sources of both indoor PM2.5 and
coarse-mode particles.
The impact of indoor source events was also observed
in I/O correlations and I/O ratios for daytime data. I/O
correlations were reduced due to the contributions of in-
door particle sources, and this demonstrates how indoor-
generated particles can mask the underlying relationship
between ambient PM concentrations and indoor concen-
trations of ambient PM, which is a central assumption of
epidemiologic studies.55 Nighttime non-source data more
accurately reflect the contributions of ambient particles
to indoor particle concentrations, and future work will
use these data to investigate ambient particle infiltration
and the factors affecting infiltration.
ACKNOWLEDGMENTS
The authors would like to express their sincere gratitude
to all of the study participants as well as to George Allen,
Jim Sullivan, Mark Davey, Peter Babich, Denise Belliveau,
and Jessica Sekula for their invaluable assistance during
the field work. In addition, we would like to thank Drs.
Eileen Abt, Paul Catalano, and Kim Thompson for their
insightful contributions during data analysis. This study
was funded by the Center for Indoor Air Research (CIAR)
under contract #96-08A. Christopher Long was supported
in part by the CIAR and the U.S. EPA STAR Graduate Fel-
lowship Program.
REFERENCES
1. Dockery, D.W.; Schwartz, J.; Spengler, J.D. Environ. Res. 1992, 59,
362–373.
2. Dockery, D.W.; Pope, C.A.; Xu, X.; Spengler, J.D.; Ware, J.H.; Fay, M.E.;
Ferris, B.G.; Speizer, F.E. N. Eng. J. Med. 1993, 329, 1753–1759.
3. Pope, C.A.; Dockery, D.W.; Schwartz, J. Inhal. Tox. 1995, 7, 1–18.
4. Schwartz, J. Environ. Res. 1994, 64, 26–35.
5. Schwartz, J.; Dockery, D.W.; Neas, L.M. J. Air & Waste Manage. Assoc.
1996, 46, 927–939.
6. Air Quality Criteria for Particulate Matter, Volumes I, II, and III; EPA/
600/P-95/001; U.S. Environmental Protection Agency; Office of Re-
search and Development: Washington, DC, 1996.
7. Abt, E.; Suh, H.H.; Allen, G.; Koutrakis, P. Environ. Health Perspect.
2000, 108, 35–44.
8. Kamens, R.; Lee, C.-T.; Weiner, R.; Leith, D. Atmos. Environ. 1991,
25A, 939–948.
9. Özkaynak, H.; Xue, J.; Spengler, J.; Wallace, L.; Pellizzari, E.; Jenkins,
P. J. Exposure Anal. Environ. Epidemiol. 1996, 6, 57–78.
10. Özkaynak, H.; Xue, J.; Weker, R.; Butler, D.; Koutrakis, P.; Spengler, J.
The Particle TEAM (PTEAM) Study: Analysis of the Data, Vol. III; U.S.
Environmental Protection Agency: Research Triangle Park, NC, 1994.
11. Raunemaa, T.; Kulmala, M.; Saari, H.; Olin, M.; Kulmala, M.H. Aerosol
Sci. Technol. 1989, 11, 11–25.
12. Tu, K.-W.; Knutson, E.O. Aerosol Sci. Technol. 1988, 9, 71–82.
13. Wallace, L.; Quakenboss, J.; Rodes, C. Continuous Measurements of
Particles, PAH, and CO in an Occupied Townhouse in Reston, VA. In
AWMA/EPA Symposium on the Measurement of Toxic and Related Air
Pollutants; Air & Waste Management Association: Research Triangle
Park, NC, 1997.
14. Highsmith, V.R.; Hardy, R.J.; Costa, D.L.; Germani, M.S. Environ. Sci.
Technol. 1992, 26, 673–680.
15. Koutrakis, P.; Briggs, S.L.K.; Leaderer, B.P. Environ. Sci. Technol. 1992,
26, 521–527.
16. Traynor, G.W.; Anthon, D.W.; Hollowell, C.D. Atmos. Environ. 1982,
16, 2979–2987.
17. Thatcher, T.L.; Layton, D.W. Atmos. Environ. 1995, 29, 1487–1497.
Cooking, cleaning, and other general activities involv-
ing either combustion (e.g., candles) or resuspension (e.g.,
any physical movement such as walking, dusting, vacu-
uming, etc.) were identified as important indoor particle
sources. In addition, as suggested by recent experimental
data for a simulated indoor environment,38 these data for
typical conditions in several study homes provide new
information that indoor chemical reactions may be an
important source of fine and ultrafine particles in indoor
residential environments. Five events for a variety of in-
door conditions demonstrated large-scale ultrafine par-
ticle formation when a commercial cleanser containing
terpenes was used. More than 50% of the particles by vol-
ume generated during these ozone/terpene events were
ultrafines. Furthermore, the PM2.5 concentration rose over
Volume 50 July 2000
Journal of the Air & Waste Management Association 1249

Long, Suh, and Koutrakis
18. Wallace, L.A. J. Air & Waste Manage. Assoc. 1996, 46, 98–126.
19. Abt, E.; Suh, H.; Catalano, P.; Koutrakis, P. Environ. Sci. Technol., in press.
20. Patashnick, H.; Rupprecht, E.G. J. Air & Waste Manage. Assoc. 1991,
41, 1079–1083.
21. Rupprecht & Patashnick Co., Inc. Operating Manual, TEOM Series
1400a Ambient Particulate (PM-10) Monitor. Rupprecht & Patashnick
Co., Inc., 1993.
43. Jang, M.; Kamens, R.M. Atmos. Environ. 1999, 33, 459–474.
44. Novakov, T.; Penner, J.E. Nature 1993, 365, 823–826.
45. White, W.H.; Macias, E.S. Aerosol Sci. Technol. 1989, 10, 111–117.
46. Rogge, W.F.; Hildemann, L.M.; Mazurek, M.A.; Cass, G.R.; Simoneit,
B.R.T. Environ. Sci. Technol. 1991, 25, 1112–1125.
47. Rogge, W.F.; Hildemann, L.M.; Mazurek, M.A.; Cass, G.R.; Simoneit,
B.R.T. Environ. Sci. Technol. 1993, 27, 2736–2744.
22. Marple, V.A.; Rubow, K.L.; Turner, W.; Spengler, J.D. J. Air Pollut. Con-
trol Assoc. 1987, 37, 1303–1307.
23. Lioy, P.J.; Wainman, T.; Turner, W.; Marple, V.A. J. Air Pollut. Control
Assoc. 1988, 38, 668–670.
48. Rogge, W.F.; Hildemann, L.M.; Mazurek, M.A.; Cass, G.R.; Simoneit,
B.R.T. Environ. Sci. Technol. 1997, 31, 2731–2737.
49. Rogge, W.F.; Hildemann, L.M.; Mazurek, M.A.; Cass, G.R.; Simoneit,
B.R.T. Environ. Sci. Technol. 1998, 32, 13–22.
24. Van Loy, M.; Saxena, P.; Babich, P.; Lawrence, J.; Allen, G.; Koutrakis,
P. J. Air & Waste Manage. Assoc., in press.
25. Chow, J.C.; Watson, J.G.; Pritchett, L.C.; Pierson, W.R.; Frazier, C.A.;
Purcell, R.G. Atmos. Environ. 1993, 27A, 1185–1201.
26. Brüel & Kjær. Instruction Manual for the Multi-gas Monitor Type 1302;
Brüel & Kjær: Denmark, 1990.
27. Lebret, E.; Boleij, J.; Brunekreef, B. Home Ventilation Under Normal
Living Conditions. In Indoor Air ’90: Proceedings of the 5th Interna-
tional Conference on Indoor Air Quality and Climate, Vol. 4: Toronto,
Canada, 1990; pp 413–418.
28. Allen, G.; Sioutas, C.; Koutrakis, P.; Reiss, R.; Lurmann, F.W.; Roberts,
P.T. J. Air & Waste Manage. Assoc. 1997, 47, 682–689.
29. Ayers, G.P.; Keywood, M.D.; Gras, J.L. Atmos. Environ. 1999, 33, 3717–
3721.
50. Delfino, R.J.; Zeiger, R.S.; Seltzer, J.M.; Street, D.H. Environ. Health
Perspect. 1998, 106, 751–761.
51. Morgan, G.; Corbett, S.; Wlodarczyk, J. Am. J. Public Health 1998, 88,
1761–1766.
52. Korrick, S.A.; Neas, L.M.; Dockery, D.W.; Gold, D.R.; Allen, G.A.; Hill,
L.B.; Kimball, K.D.; Rosner, B.A.; Speizer, F.E. Environ. Health Perspect.
1998, 106, 93–99.
53. Gold, D.R.; Litonjua, A.; Schwartz, J.; Lovett, E.; Larson, A.; Nearing,
B.; Allen, G.; Verrier, M.; Cherry, R.; Verrier, R. Circulation, in press.
54. Sarnat, J.A.; Koutrakis, P.; Suh, H.H. J. Air & Waste Manage. Assoc., in
press.
55. Mage, D.; Wilson, W.; Hasselblad, V.; Grant, L. J. Air & Waste Manage.
Assoc. 1999, 49, 1280–1291.
30. Sioutas, C.; Abt, E.; Wolfson, J.M.; Koutrakis, P. Aerosol Sci. Technol.
1999, 30, 84–92.
31. Peters, T.M.; Chein, H.; Lundgren, D.A.; Keady, P.B. Aerosol Sci. Technol.
1993, 19, 396–405.
32. Peters, T.M.; Vanderpool, R.; Burton, R.M.; Koutrakis, P.; Abt, E.;
Lawrence, J. Combination of Aerodynamic Particle Sizer and Scan-
ning Mobility Particle Sizer Data in Measuring Ambient Aerosols. In
Fourteenth Annual Meeting of the American Association for Aerosol Re-
search, Vol. 370; American Association for Aerosol Research: Pitts-
burgh, PA, 1996.
33. Hildemann, L.M.; Markowski, G.R.; Jones, M.C.; Cass, G.R. Aerosol
Sci. Technol. 1991, 14, 138–152.
34. Löfroth, G.; Stensman, C.; Brandhorst-Satzkorn, M. Mutat. Res. 1991,
261, 21–28.
About the Authors
Christopher M. Long is a doctoral student in the Environ-
mental Science and Engineering Program at the Harvard
School of Public Health in Boston, MA. Dr. Helen Suh is an
assistant professor of environmental chemistry and expo-
sure assessment at the Harvard School of Public Health.
Dr. Petros Koutrakis is a professor of environmental sci-
ences at the Harvard School of Public Health. Correspon-
dence should be addressed to Christopher M. Long, 665
Huntington Avenue, Building I, Room B-21, Boston, MA
02115 (clong@hsph.harvard.edu).
35. Kildesø, J.; Vallarino, J.; Spengler, J.D.; Brightman, H.S.; Schneider, T.
Atmos. Environ. 1999, 33, 699–707.
36. Rodes, C.E.; Kamens, R.M.; Wiener, R.W. Indoor Air 1991, 2, 123–145.
37. Lewis, R.G.; Fortune, C.R.; Willis, R.D.; Camann, D.E.; Antley, J.T.
Environ. Health Perspect. 1999, 107, 721–726.
38. Weschler, C.J.; Shields, H.C. Atmos. Environ. 1999, 33, 2301–2312.
39. Finlayson-Pitts, B.J.; Pitts, J.N., Jr. Atmospheric Chemistry: Fundamen-
tals and Experimental Techniques; John Wiley & Sons: New York, 1986.
40. Calogirou, A.; Larsen, B.R.; Kotzias, D. Atmos. Environ. 1999, 33, 1423–
1439.
41. Yu, J.; Griffin, R.J.; Crocker, D.R.; Flagan, R.C.; Seinfeld, J.H.; Blanchard,
P. Geophys. Res. Lett. 1999, 26, 1145.
42. Kavouras, I.G.; Mihalopoulos, N.; Stephanou, E. Environ. Sci. Technol.
1999, 33, 1028–1037.
1250 Journal of the Air & Waste Management Association
Volume 50 July 2000