Long duration tests of room air filters in cigarette smokers' homes.

Abstract

Information regarding the long-term performance of stand-alone room airfilters is limited. In this study, laboratory and field tests were carried out to determine the effectiveness and performance of room filters that are easily deployed in essentially any type of house. Tests were conducted in houses containing strong PM sources, specifically cigarette smokers. Using commercially available four-speed HEPA filter units, we tested flow rate, pressure drop, and power consumption as a function of fan speed and filter loading. Filters were then deployed in four single-family homes over a 2 month period. Between 15 and 40 cigarettes were smoked daily by several smokers in each home. Occupants were instructed to continuously operate the unit at one of the higher speeds. Periodically, we monitored filter usage, fan speed, particulate matter (PM) mass concentrations, PM number concentrations, volatile organic compound (VOC) levels, and other parameters with the filter fan operating and with filters both installed and removed. The filters decreased PM concentrations by 30-70%, depending on size fraction and occupant activities, and significantly reduced the half-life of PM3-1.0. The half-life of 1-5 microm particles, CO2 concentrations, and VOC concentrations, including 2,5-dimethyl furan (a tracer for environmental tobacco smoke), did not change, indicating that occupancy and cigarette smoking intensity did not change overthe monitoring periods. Occupants generally kept the filters operating at a moderate speed. Filter air flow rates decreased 7-14% with extended operation, largely due to the loading of prefilters. Air exchange rates, deposition loss rates, and clean air delivery rates were estimated from the field data. Continuous operation at an intermediate fan speed would incur a total annualized cost of $236. While acceptance of the filters was very high, occupants might benefit from instructions and reminders to clean the prefilter and to keep the unit on. We conclude that adequately sized room air filters can substantially lower PM concentrations in smoker's homes if air exchange rates are limited and that the filters can maintain their performance over extended periods.

Full text

Page 1

Long Duration Tests of Room Air
Filters in Cigarette Smokers’ Homes
S T U A R T B A T T E R M A N , *
C H R I S T O P H E R G O D W I N , A N D
C H U N R O N G J I A
Environmental Health Sciences, University of Michigan,
Ann Arbor, Michigan 48109-2029
Information regarding the long-term performance of
stand-aloneroomairfiltersislimited.Inthisstudy,laboratory
and field tests were carried out to determine the
effectiveness and performance of room filters that are
easily deployed in essentially any type of house. Tests were
conducted in houses containing strong PM sources,
specificallycigarettesmokers.Usingcommerciallyavailable
four-speed HEPA filter units, we tested flow rate, pressure
drop, and power consumption as a function of fan
speed and filter loading. Filters were then deployed in
four single-family homes over a 2 month period. Between
15 and 40 cigarettes were smoked daily by several
smokers in each home. Occupants were instructed to
continuously operate the unit at one of the higher speeds.
Periodically, we monitored filter usage, fan speed,
particulate matter (PM) mass concentrations, PM number
concentrations, volatile organic compound (VOC) levels,
and other parameters with the filter fan operating and with
filters both installed and removed. The filters decreased
PM concentrations by 30-70%, depending on size fraction
and occupant activities, and significantly reduced the half-
life of PM0.3-1.0. The half-life of 1-5 µm particles, CO2
concentrations, and VOC concentrations, including 2,5-
dimethyl furan (a tracer for environmental tobacco smoke),
did not change, indicating that occupancy and cigarette
smokingintensitydidnotchangeoverthemonitoringperiods.
Occupants generally kept the filters operating at a
moderate speed. Filter air flow rates decreased 7-14%
withextendedoperation,largelyduetotheloadingofprefilters.
Air exchange rates, deposition loss rates, and clean air
deliveryrateswereestimatedfromthefielddata.Continuous
operation at an intermediate fan speed would incur a
total annualized cost of $236. While acceptance of the filters
was very high, occupants might benefit from instructions
and reminders to clean the prefilter and to keep the unit on.
We conclude that adequately sized room air filters can
substantially lower PM concentrations in smoker’s homes
if air exchange rates are limited and that the filters can
maintain their performance over extended periods.
Introduction
In-homefilters,sometimescalledairpurifiersorroomfilters,
are designed to reduce concentrations of particulate matter
(PM) such as tobacco smoke, pollen, dust mites, animal
allergens,anddieselexhaustparticles.Theuseofsuchfilters
might help to ameliorate respiratory problems such as
asthma, which affects a large and growing portion of the
population,especiallychildrenlivinginpovertyandinurban
centers(1,2).Informationregardingtheperformanceoffilters
inresidentialsettingsislimited,andvirtuallynoinformation
exists regarding their long-term performance. Further, the
literature is mixed regarding the effectiveness of filters in
managing asthma and other respiratory problems, likely a
resultofpoorlycontrolledtests.Inthisstudy,laboratoryand
field tests were conducted to determine the effectiveness of
stand-alone room filters. Extended tests in households
containing strong PM sources, specifically smokers, were
emphasized.
Most American homes have forced air heating systems
(e.g., 81% in Detroit MI (3)), most often an inexpensive and
disposable fiberglass furnace filter. These filters function
mainly to protect the fan and furnace heat exchanger, and
no claims are made regarding particle removal efficiencies.
Particleremovalefficiencyforthesefiltersispoor(e.g.,<20%
for 0.02-2 µm diameter particles and <5% for 0.1-0.5 µm
diameter particles (4)). Much more efficient filters are
available, both as new equipment and as upgrades for
conventionalfurnacefilters(e.g.,pleatedmedium-efficiency
filters, electrostatically charged filters, electrostatic precipita-
tors,andhigh-efficiencyparticlearresting(HEPA)filtersrated
toremove>99.97%ofparticles>0.3µmindiameter).Houses
using space heaters, radiators, and baseboard heating
typically employ no filters. Room air filters may be used in
such houses, as well as in houses with forced air systems, to
provide additional filtration. These units typically employ
medium-to-high efficiency paper filters, although ionizers
and water-washing units are also marketed.
Laboratory and Theoretical Evaluations. Performance
evaluationsoffiltershaveemphasizedlaboratory(chamber)
tests and model simulations. Removal efficiency, often
expressed as the clean air delivery rate (CADR), the product
ofthefilter’sairflowrateandefficiencyforachallengeaerosol,
maybemeasuredbysingle-passchallengetechniquesusing
the filter element alone (5) and, more commonly, using
chamber or test house studies (6-8). Evaluations of room
filters have been completed for many physical parameters
(e.g., dust, spores, environmental tobacco smoke (ETS), and
nicotine) as well as sensory measures (e.g., odor strength,
nasal irritation, and eye irritation (9-14)).
Theoreticalmodelingandchamberstudiesshowthatthe
key parameters affecting performance are the particle size
distribution, filtration efficiency, mixing and capture ef-
ficiency, and most importantly, filter air flow. Air flow rates
through room filters must be equivalent to at least several
air exchanges per hour to obtain substantial (>50%) PM
reductions (12, 13, 15). Filter efficiencies above 85% provide
only modest gains in performance (15). While in-duct filters
are generally more effective than stand-alone units (4, 16-
18), such units cannot be used in homes without forced air
systems. Additionally, installation and operating costs and
space issues may prohibit their use. Performance criteria
such as CADR may not reflect in-use performance due to
differences in ventilation, air flow patterns, PM source
locations,PMdispersalcharacteristics,PMsizedistributions,
and other factors. Typically both modeling and chamber
studiesassumewell-mixedconditions,whichareunlikelyin
practice (e.g., some short-circuiting of air will occur due to
the proximity of the filter’s inlet and outlet (6)). PM in the
breathing zone reflects both generation and removal pro-
cesses, and PM reductions in the breathing zone are never
as significant as those achieved at the filter (19).
* Correspondingauthorphone: (734)763-2417;fax: (734)936-7283;
e-mail: StuartB@umich.edu.

Page 2

The performance of filters appears to degrade over time,
although most information on this topic is anecdotal. For
example, the CADR for a HEPA unit installed in a residential
bedroom decreased by 25% after intermittent operation for
800h(11).Asdustloadingsincrease,filtrationefficiencyand
CADR suffer on electrostatically charged filters; however,
filtration efficiencies may increase for other filter types at
the cost of increased pressure drop and decreased air flow
andCADR(4,20).Dirtonelectrostaticprecipitatorshasbeen
noted to reduce removal efficiency (11).
FieldEvaluations.Informationregardingtheeffectiveness
of filters in the field is scarce. Several studies, primarily in
houses, have examined the effect of filters on particle mass
and number concentrations, but these studies did not
distinguish between particles from different sources (15).
Most studies have focused on reductions in large particles
associated with animal allergens, pollen, and dust mites (9,
12, 13, 21), but analyses may have been confounded by
temporalchanges,thelackofcontrols,inadequatestudysizes,
and short-duration tests. For these reasons, the literature is
mixed on the effectiveness of air filters. A HEPA room filter
in a 24 h test removed 80-90% of pollen, fungal spores, and
particles in a bedroom, as compared to ∼50% removed due
to sedimentation in a comparable control bedroom without
a filter (12). In contrast, no difference was seen between
baseline (pre-intervention) and monthly post-intervention
24 h PM2.5 levels in nine apartments equipped with three
filtertypes(ionization,HEPA/O3/charcoal,andHEPAtypes)
(22). The continued use of filters in these apartments was
not confirmed, a single pre-intervention measurement was
taken, and the analysis may have been affected by window
and door opening and temporal changes in indoor sources
and outdoor pollutant levels over the 7 month study. In 7 h
tests of nine homes with dogs, HEPA-type room filters
significantly reduced concentrations of an airborne dog
allergen, Can f1, by 75% (dog in room tested) to 90% (dog
in other rooms) when windows and doors were closed (13).
Several types of commercial air filters tested in a 14 ft × 20
ft (4.3 m × 6.1 m) smoking lounge, and two filters, including
a 750 cfm (0.35 m3s-1) HEPA filter, did substantially (>50%)
reduce nicotine and PM10 levels as compared to dilution
ventilation; no ventilation rates were provided in the two 8
h tests completed (23).
Several reports of children with asthma or persistent
allergicrhinitisshowedmarkedimprovementsofsymptoms
and quality of life with air filters installed in bedrooms (10,
24). HEPA and air filtering devices have some documented
benefit in reducing symptoms in allergic asthmatics (25),
although these results are not always reproducible (26, 27).
None of these studies include the simultaneous use of air
conditioners that are needed to limit air exchange in warm
weather or otherwise control for air exchange rates, which
can increase tremendously as when windows are opened.
Experimental Procedures
WetestedconventionalHEPAtyperoomfiltersinlaboratory
and field tests designed to be representative of households
in the Detroit, MI area.
Air Filter. The selected stand-alone room air filter
(WhirlpoolWhispure510,28)wasrecentlyhighlyranked(29).
This 61 cm × 52 cm × 25 cm unit has a carbon-impregnated
prefilter,apleatedHEPA-typefilter,fourfanspeeds,indicator
lights for fan speed and replacement times for filters, and a
clean air delivery rating of 330 cfm (0.156 m3s-1, ref 28). The
manufacturerspecifiesreplacementoftheprefilterandHEPA
filtersevery3and12months,respectively.Acurrent-sensing
circuitcoupledtoadatarecorderwasaddedtomonitorusage
andfanspeed,alongwithindoortemperatureandhumidity
(Hobo H08, Onset Computer Corporation, Bourne, MA). To
measurepressuredropacrossthefilter,weinsertedapressure
sensing tube through a small (1.6 mm) hole. These changes
did not affect the unit’s operation.
LaboratoryTests.Thefilterunit’sairflow,pressuredrop,
and electricity consumption were measured at multiple fan
speeds and filter loadings. Filter loadings were simulated by
blocking 0, 25, 50, 75, and 100% of the filter. At each loading,
thefilterpressuredropwasmeasuredusingaprecisiongauge
(Baratron, PDR-C-1B, MKS Instruments, Inc., Burlington,
MA),andairflowwasmeasuredusinganintegratingvelocity
sensor (Velogrid, Airdata Multimeter ADM-870, Shortridge
Instruments,Inc.,Scottsdale,AZ).Fancurvesweregenerated
foreachspeedsetting.Twoidenticalunitsweretested,initially
when new and after several months in smoking households.
Field Tests. Field tests were carried out in four Detroit-
area houses containing smokers. The participating head of
household provided informed and written consent, in
accordance with Institutional Review Board Procedure
guidelines and approvals from the University of Michigan.
Participantswereprovidedwithamodestcashincentiveand
compensated for the electricity consumed. A walkthrough
checklist noted room size, location of vents and windows,
and other factors. Participants completed a survey about
PM-emitting activities in the home, including the frequency
of cigarette smoking, cooking, and cleaning activities. The
followingsingle-familyhouseswerestudiedoverMarch,April
and May:
(i)single-levelmobilehome,12yearsoldwithtwosmoking
adults, two nonsmoking children and a small dog. A total of
20 cigarettes was reported to be smoked in the home on
weekdays, 15 in the room with the air filter, and 40 on
weekends.Thehomewaslocatedinaruralareaalongabusy
state highway. There was no garage. Heating and cooling
were provided by a gas-fired forced air furnace/central air
conditioner in a service closet on the main floor. Cooking
usedelectricapplianceswitharecirculatingstovehood.The
sampling area was in the 10 ft × 20 ft × 8 ft (45.3 m3) living
and kitchen/dining area of the open floorplan home; this
area contained the front door and one window, which was
not opened during the study. The living room floor was
carpeted, and the kitchen/dining area had sheet flooring.
The home was clean with some clutter (mostly toys and
childcare items).
(ii) Three-floor split-level house, 23 years old with two
smokingadultresidentsandtwocats,locatedinaresidential
neighborhood two blocks from a busy five-lane main
highway. A total of 15 cigarettes was reported to be smoked
inthehomeonweekdays(allintheroomwiththefilter),and
20 on weekends. The home had an attached garage where
twocarswereusuallyparkedwhentheresidentswerehome.
Heating and cooling were provided by a gas-fired forced air
furnace/central air conditioner fitted with a humidifier.
Cooking used electric appliances with a recirculating stove
hood. The study space was located in the carpeted living
room of the 10 ft × 12 ft × 8 ft (27.2 m3) finished basement,
whichhadonewindowthatwasnotopenedduringthestudy.
The home was clean; however, the kitchen floor was being
renovated with tiles. Other floors were wood or carpet.
(iii) Single-floor home, 50 years old with two smoking
adultresidents,twononsmokingchildren,andtwomedium-
sizeddogsthatremainedoutside.Atotalof35cigaretteswas
reported to be smoked daily in the home by occupants and
visitors during the week, 25 in the room with the filter, and
25 in the home on weekends. The home was in a residential
neighborhoodoneblockfromabusymainthoroughfare.An
unattached garage was used for storage, and one car was
parked adjacent to the living room windows. Heating was
provided by a free-standing gas-fired space heater (Boston
stove) in the living room and vented outdoors by a flue pipe
passingthroughtheceilingandroof.Cookingwasdoneusing
a gas range without a hood, and the occupants reported

Page 3

frequentfrying(threetofivetimesperweek).Thestudyspace
was in the carpeted 15 ft × 12 ft × 8 ft (40.7 m3) dining area,
located between the kitchen and the living room. This area
had two windows that were not opened during the study.
Theflooringinlivinganddiningroomswascarpet;thekitchen
had sheet flooring. There was evidence of water damage to
thekitchenceiling.Thehomewasrelativelycleanandclutter-
free.
(iv) Single-floor home, 50 years old with two smoking
adult residents, one nonsmoking teenager, and a small dog.
A total of 20 cigarettes was reported to be smoked daily in
thehousebyoccupantsandvisitors,18intheroomwiththe
filter,and30inthehomeonweekends.Thehomewaslocated
in a residential neighborhood 1.5 km from two busy five-
lanemainhighwaysandaheavilytraveledexpressway.There
was no garage or storage shed: cars were parked on the
street.Heatingwasprovidedbyagas-firedforcedairfurnace
in the unfinished basement that had evidence of recent
malfunction (e.g., soot inside the enclosure and outside of
the burner box). The owner reported having the furnace
recently serviced for this problem. Cooking used a gas range
without a hood. The study space was located in the 21 ft ×
11 ft × 7.5 ft (49 m3) carpeted living/dining area located
between the front door and the kitchen that had three
windowsthatwerenotopenedduringthestudy.Thekitchen
had sheet flooring. The home was clean and clutter-free.
Laundry facilities were in the basement; a dryer was vented
indoors and may have been the source of the slight water/
condensation damage noted on several windows and the
slight musty odor in the basement. There was no evidence
of an obvious mold infestation.
In each house, the filter unit was placed in the central
livingarea,andresidentswereinstructedtokeeptheuniton
continuously at one of the higher speeds, subject to noise
concerns.Theunit’susagewascontinuouslymonitoredand
recorded. Air quality monitoring and filter efficiency tests
were made in the week when the purifier was installed and
the week before it was removed. Measurements were taken
in the houses both in parallel and on subsequent weeks. In
both cases, the study design utilized each house as its own
control by taking initial 2 day baseline measurements early
ineachstudyweek(filterremoved)andimmediatelyfollowed
by 2 day measurements as the test case (filter installed). In
this manner, seasonal and spatial changes become less
important.
The field tests included the following continuous mea-
surements: opticalparticlenumbercountsin0.3-1and1-5
µm diameter size ranges, recorded every 1 min and reduced
to5minaverages(GT-521,MetOne,GrantsPass,OR);carbon
dioxide (CO2) concentrations recorded every 5 min (GMW-
21, Vasaila, Helsinki, Finland); carbon monoxide (CO)
recordedevery5min(H11-001,OnsetComputerCorporation,
Bourne, MA); and temperature and relative humidity re-
cordedevery5min(HoboH08,OnsetComputerCorporation,
Bourne, MA). The optical instruments were new and had
been recently calibrated by the manufacturer. The optical
measurements include only the larger, secondary combus-
tion-related particles (primary combustion particles in
tobacco smoke have a <0.3 µm diameter), but they should
closely track total particle generation and removal by the
filter.Complementaryparticlemeasurementsincluded2day
PM open-face filter samples, weighed using a microbalance
(Model 29, Cahn Instruments, Inc., Cerritos, CA) after
temperature and humidity conditioning of the filter. These
time-integrated samples represent total suspended particu-
late (TSP) samples as they capture essentially the full size
range of airborne particles. Additionally, volatile organic
compound(VOC)concentrationsoverthesame2dayperiods
were determined by active sampling onto adsorbent tubes
containing Tenax-GR and Carbosieve SIII, followed by
thermal desorption, cryofocusing, and GC-MS analysis for
95 compounds (30).
Homes were selected specifically due to the presence of
ETS; thus, it was necessary to determine whether ETS levels
changed over the monitoring period. ETS levels have been
estimated using ultraviolet absorbance and fluorescence of
the methanol extract of PM, solanesol, scopoletin, vapor-
phase nicotine (VPN), and other tracers (31). While VPN has
been the most widely used tracer for ETS (32), its suitability
has been questioned because it exhibits sink/source, decay,
and UV radiation effects that differ from other ETS con-
stituents (e.g., much of the VPN is deposited to surfaces,
while most ETS is removed by ventilation); thus, ratios of
ETS (measured as RSP or PM2.5) to nicotine can vary
substantiallywithtimeandventilationrates(33,34).Recently,
2,5-dimethyl furan has been identified as a unique vapor-
phase tracer of ETS (35). Our measurements in smokers’
homes have shown 2,5-dimethyl furan concentrations present
at low (0.6-1 µg m-3) levels but easily quantifiable given
detection limits of ∼0.01 µg m-3. We do not detect this VOC
in nonsmoking environments. Hence, this VOC was used as
a sensitive and selective tracer for ETS.
Data Analyses and Modeling. Analyses included t-test
comparisons of pollutant concentrations over the periods
with and without the filter installed. Changes in CO2helped
to show whether the occupancy changed, and changes in
2,5-dimethyl furan showed whether the ETS levels changed.
Ifnot,thenchangesinPMconcentrationsshowtheinfluence
of the air filter.
To determine the particle removal achieved by the filter,
concentrationdecaycurvesthatfollowedanemissionevent
(e.g., smoking) were fit to a first-order model
where Ct) particle number concentrations (#/L); t ) time
(h);C0)initialconcentration;Ke)decayrate(h-1);andCmin
) background or minimum concentration. Parameters C0,
Ke, and Cminwere estimated by maximum likelihood using
logarithmsofconcentrationsformultipleandisolatedevents
(e.g., the last smoking event of the day or night, allowing a
long(3-7h)andwell-defineddecaycurve).Parameterswere
estimatedforfineandcoarseparticles(<1µmdiameterand
1-5 µm diameter, respectively) with the filter operating,
giving the experimental decay rate Ke, and without the filter,
giving a natural decay rate Kn attributable to particle
deposition on surfaces in the house and removal via air
exchange. We also estimated decay curves using CO2
concentrations following a change in occupancy, giving an
air exchange rate Ka (h-1). (CO2 deposition and removal
processesarenegligibleonthetimescaleconsidered.)When
we could estimate Ka, no difference was seen in Kabetween
periods with and without the filter in place, as expected. Ka
was not estimated if well-defined CO2peaks or decay curves
were not seen (e.g., a result of occupancy being more or less
constant).Fitsbetweeneq1andobservedconcentrationsof
PM0.3-1and CO2were very high (e.g., typically r >0.97). Ke,
Kn, and Kawere estimated for several well-defined events in
each house at the beginning and end of the study. The
pollutant half-life t1/2was computed as t1/2) ln(2)/K.
In chamber tests where the chamber’s volume V (m3) is
known and constant, the clean air delivery rate, CADR (m3
s-1), is computed as (11)
Inhouses,thesameapproachmaybeused,buttheeffective
volume may be uncertain since mixing/exchange occurs
betweensomebutpossiblynotallrooms.Particledeposition
Ct) C0exp(-Ket) + Cmin
(1)
CADR ) V(Ke -Kn) (2)

Page 4

velocity Vd(cm s-1) is estimated as
whereKpisthedepositionlossrate,thenaturalparticledecay
occurring without the filter that accounts for air exchange;
thus Kp) Kn- Ka, and S/V is the surface-to-volume ratio of
theroomorhouse,whichrangesfrom2to3m-1infurnished
homes and offices (4, 6, 15).
Indoor PM concentrations are usually predicted using
fully mixed models that account for infiltration, makeup air,
particle deposition, indoor and outdoor sources, and recir-
culationofairthroughtheroomairfilter(15,36).Weassume
that resuspension, coagulation, and reaction are negligible
and,also,thatparametervariationisuncorrelated.Following
Riley et al. (36), the steady-state indoor concentration is
where C and Co are indoor and outdoor concentrations,
respectively (µg m-3); p is the particle penetration factor; Qi,
Qn, Qm, and Qr are infiltration, natural, makeup, and
(recirculating) air filter air flow rates, respectively (m3s-1);
V is the room or house volume (m3); S is emission rate of
indoor sources (µg s-1); ηm and ηr are filter efficiency for
makeupandrecirculating(room)filters,respectively;andKp
is the deposition loss rate coefficient (s-1). Because we are
mostinterestedintheeffectofthefilter,wederivetherelative
reductioninPMconcentrations,R,achievedbytheoperating
filter as
whereCfandCnareconcentrationsachievedwithandwithout
thefilteroperating,respectively,andAERistheairexchange
rate ) (Qi+ Qm+ Qn)/V (s-1). Conveniently, this leaves only
five terms: Qr, V, AER, ηr, and Kp. The latter two depend on
particlecharacteristics,mostnotablysize.Eq5indicatesthat
the relative reductions in PM concentrations achieved by a
room air filter do not depend on the strength or level of
indoor or outdoor sources and the efficiency of makeup
filtration, if any (rare in U.S. houses), and that Qrηr/V must
be large relative to Kp + AER to provide substantially
decreased concentrations.
Results
LaboratoryTests.Thetwofilterunitstestedinthelaboratory
had very similar performances when new. Air flow rates
ranged from 235 to 439 cfm (0.111-0.207 m3s-1), pressure
drops across the filter were from 0.37 to 0.82 mmHg, and
current draw was 0.54-1.0 A, all depending on fan speed
(Table1).Fancurveswererelativelyflatasafunctionoffilter
loading,indicatingthattheunit’selectricitydemandchanged
little as a function of loading (Figure 1). The fan speed (and
overalluse)waseasilydeterminedbythecurrentmonitoring
device. Flow rates and pressure drops changed moderately
as the filter was obstructed (e.g., 50% blocking decreased
flows by 15-21% and increased pressure drop by 26-55%),
depending on fan speed.
After 2 months of continuous operation in the smokers’
homes, the prefilter had collected a heavy coating of fibrous
graydust,andtheHEPAfilter’supstreamsidehaddarkened
slightly. The downstream side remained white. Both filters
smelled of cigarette smoke. Air flows dropped by 7-14%,
and pressure drops across the filters had increased by an
average of 7-16%, depending on fan speed and unit (Table
2). (The unexpected pressure decrease at the low speed is
due to the limited precision of the instrumentation.) On the
basisofacomparisonofpressuredropmeasuredacrossused
filters versus our simulated loading of new filters, 2 months
ofusewasequivalenttoblocking33%ofthefilterarea.Most
of the additional pressure drop was attributable to the filter
cakeontheprimaryfilter.Vacuuming,washing,orreplacing
the primary filter restored flows to within 5% of new
conditions (Table 2). It should be noted that the filter unit
has lights indicating the need for filter replacement, and the
prefilter light had been activated. Despite the increased
pressure drop across the prefilter, the HEPA filter caused
most (81-89%, increasing with fan speed) of the pressure
drop in the system.
Field Tests. Initial tests at the four houses showed some
house-to-house and month-to-month variation in the PM
concentrations, particle counts, 2,5-dimethyl furan concen-
trations (the ETS tracer), and CO2concentrations (Table 3).
However, 2 day average levels of cigarette smoke generated
within each house did not change significantly during the
weeks monitored or depend on whether the HEPA filter was
Vd) Kp/(S/V) (3)
C ) -CopQi+ CoQn+ CoQm(1 - ηm) + S
Qrηr+ KpV + Qm+ Qi+ Qn
(4)
R ) 1 -
Cf
Cn
) 1 -
Kp+ AER
Qrηr/V + Kp+ AER
(5)
TABLE 1. Tests of New Units Showing Filter Air Flow, Current,
and Filter Pressure Drop at Four Fan Speeds and Five Loading
Conditionsa
loading
(%)
fan
speed
air
flow (cfm)
current
(amps)
pressure
drop (mmHg)
0 low
medium
high
turbo
low
medium
high
turbo
low
medium
high
turbo
low
medium
high
turbo
low
medium
high
turbo
235
298
340
429
221
281
320
404
201
254
269
356
167
210
248
281
0.538
0.640
0.755
0.997
0.540
0.641
0.748
0.987
0.521
0.620
0.720
0.960
0.507
0.600
0.710
0.930
0.460
0.540
0.640
0.865
0.37
0.52
0.72
0.82
0.47
0.63
0.97
0.98
0.57
0.72
0.90
1.05
0.67
0.85
1.00
1.28
1.10
1.37
1.50
1.80
25
50
75
1000
0
13
14
aAverage of two units, three replicates each.
FIGURE 1. Fan curves for room air cleaner at four speeds and
loadings from 0 to 100% of filter area. Average of two units, three
replicates at each loading.

Page 5

installed, based on the PM number concentration plots,
whichshowedaboutthesamenumberofpeaks(typically10
smoking events daily), and by the 2,5-dimethyl furan
concentration,whichchangedonlyslightlyduringtheweeks
investigated.Occupancyandventilationwerealsounchanged
on a 2 day basis, as indicated by CO2 concentrations and
occupant surveys. Concentrations of toluene (Table 3) and
other VOCs (e.g., benzene, styrene, R-pinene, and D-li-
monene)exceededlevelswetypicallyobserveinnonsmoking
homes.VOCconcentrationsalsodidnotchangemuchduring
the week. All evidence indicates that pollutant generation,
occupant activity, and ventilation changed little within the
initial and final weeks of study.
The effect of operating filters was dramatic. PM mass
concentrations dropped from 86 ( 43 µg m-3without the
filter to 53 ( 24 µg m-3with the HEPA filter, a 38% reduction
(Table 3). Monitoring at three houses during periods when
windows were closed showed a 65% decrease in the median
number concentrations of PM0.3-1and PM1-5(Table 4). For
PM1-5, the relative reduction due to the filters (65%) applied
to both the median and the 90th percentile concentration.
For PM0.3-1, peak and 90th percentile levels did not change,
a result of instrument limitations. Although measured peak
PM0.3-1concentrations changed little with the filter in place,
the duration of peaks was considerably shorter (Figure 2).
No evidence of VOC removal by the filter was seen. The
prefilter weighed 48 g, and the amount of activated carbon
impregnated was apparently too small to cause measurable
differences in VOC concentrations. (Table 3 shows toluene
and 2,5-dimethylfuran; other VOCs also did not change
significantly.)
Smoking events saturated the particle counter at its
maximum response (7 × 105#/L), noted by a plateau or
rounded peak in the PM0.3-1 measurements that lasted
typically 5-20 min (Figure 2). In other studies using particle
mobility analyzers, smoking has produced PM0.02-1concen-
trations of ∼5 × 107#/L, with roughly 30% of the particles
(∼1.5 × 107#/L) in the optical PM0.3-1range measured here
(37, 38). In comparison, PM0.3-1 number concentrations
measured during smoking events in the houses were ∼20
timeslower.ThismaybearesultofthewidedispersalofPM
throughout the room and the house, especially given the
mixing facilitated by the velocity of the air filter’s discharge
and by the movement and other activities of smokers in the
houses. Additionally, we took 5 min averages that may have
missed peak levels. However, saturation of the optical
counters,whichclearlyoccurred,isthemainreasonforthese
low PM0.3-1concentrations. This biased the 90th percentile
TABLE 2. Tests of Filter Air Flow and Filter Pressure Drop at Four Fan Speeds When New, After 2 Months of Use, and After 2
Months of Use with HEPA Filter Onlya
new unit used unitused -HEPA only
mean
used - cleaned prefilter
mean
variable/
fan speedmean SDmean SD SDSD
Air flow rate (cfm)
9
4
7
13
Pressure drop (mmHg)
0.04
0.11
0.04
0.04
Low
Med
High
Turbo
235
298
354
429
5
5
7
3
215
269
320
378
242
302
366
427
6 231
287
343
418
6
6
4
9
10
8
3
Low
Med
High
Turbo
0.46
0.52
0.65
0.82
0.19
0.03
0.00
0.03
0.43
0.58
0.73
0.98
0.35
0.47
0.65
0.85
0.07
0.04
0.07
0.07
0.35
0.50
0.63
0.80
0.05
0.00
0.05
0.00
aBased on two units, three replicates each.
TABLE 3. Summary of Monitoring Results at Four Sites with and without Filters and with Used Filtersa
PM0. 3 - 1
mean
(#/L)
PM1 - 5
mean
(#/L)
PM-g ravim etric
mean
(µg/m3)
CO2Toluene
mean
(µg/m3)
2-5 DMF
mean
(µg/m3)
condition
SD
(#/L)
SD
(#/L)
SD
(µg/m3)
mean
(ppm)
SD
(ppm)
SD
(µg/m3)
SD
(µg/m3)
newwithout filter
with filter
without filter
with filter
without filter
with filter
249 643
116 203
332 490
199 646
291 067
157 925
26 725
21 639
171 353
128 139
110 965
89 164
1622
451
2329
1155
1976
803
934
150
1929
801
1303
622
92
49
81
57
86
53
42
27
61
31
43
24
661
723
681
806
671
764
48 29
26
29
33
29
28
30
15
32
36
27
20
0.80
1.09
0.64
0.60
0.76
0.99
0.11
0.59
0.15
0.14
0.12
0.55
111
171
109
103
102
used
all
aBased on 2 day samples for each condition. 2-5 DMF is 2,5-dimethyl furan.
TABLE 4. Percentiles (50th, 75th, 90th) of Particle Number Concentrations Measured at Three Houses with and without Filter
during Periods When Windows Were Closeda
site 2
75th
site 3
75th
site 4
75th
average
75th
PM condition50th
90th 50th 90th50th90th 50th 90th
PM0.3 1
no filter (1000 #/L)
filter (1000 #/L)
change (%)
no filter (1000 #/L)
filter (1000 #/L)
change (%)
145
52
64
992
189
81
421
172
59
2185
458
604
373
38
3714
854
507
259
49
2223
369
660
488
26
3761
749
689
581
16
9385
2578
166
32
81
562
370
34
475
104
78
1314
652
667
326
51
2185
1098
273
114
58
1259
309
519
255
51
2420
620
653
426
35
5095
1510
PM1-5
7977 83 80 735050 6670 66
aUpper percentiles of PM0.3-1are biased due to instrument limitations (see text).

Page 6

PM0.3-1levels and the corresponding percentage changes in
Table4;however,medianandmost75thpercentilestatistics
were unaffected since peak levels were short-lived. PM1-5
measurements were unaffected.
After 2 months of use, the weather had warmed consid-
erably, and the final week of tests was unseasonably hot and
humid. Several participants had opened windows, greatly
increasing ventilation and air exchange rates (39). PM levels
were generally lower. At site 4, occupants smoked and
occasionally slept on the porch, and we did not detect 2,5-
dimethylfuran, the ETS tracer, in this home. As before, all
PM indicators (filter concentrations, PM0.3-1, PM1-5) de-
creased with operating filters, although less dramatically as
compared to the earlier trial conducted in cold weather.
Table5summarizesthefirst-orderdecayanalysis.Forall
housesandseasons,thePM0.3-1half-lifefellfrom0.57(0.42
h without the filter to 0.30 ( 0.15 h with the filter. The CO2
half-life air averaged 1.8 ( 0.9 h. While concentrations of
PM1-5decreasedsignificantlywiththefilteroperating,PM1-5
half-lifeestimateswereinconsistent,trendsshowedfewpeaks
from which the half-life could be estimated, and PM1-5and
PM0.3-1peaks were not coincident, indicating that the two
size fractions had different sources. The strongest PM0.3-1
sources were primarily indoors and occurred as distinct
events (e.g., smoking and cooking). In contrast, PM1-5may
have originated from resuspension of indoor and outdoor
dust in different events that were dispersed in space and
time.
Estimates of AER and Vdvaried from house-to-house. As
mentioned, well-defined decay trends for CO2were needed
to provide consistent estimates of the AER, and variability
resulted when measured CO2 levels did not show strong
peaks. This problem would be avoided and uncertainties
reduced with the use of SF6tracer gas decay rate measure-
ments. Other parameters (Ka, Kn) affecting Vdshowed still
greatervariability,mostlyduetohouse-to-housedifferences,
and to a lesser extent due to variability in the repeated
measurements collected at each house. This variability is
expected given differences in house and emission source
configurations,roomvolumes,etc.Usingaveragevalues(and
standard deviations) across the four homes for PM0.3-1(Ka
) 0.45 ( 0.20 h-1; Kn) 1.7 ( 1.1 h-1; and Ke) 2.6 ( 1.1 h-1)
and assuming S/V ) 2 m-1, Vd) 0.018 ( 0.015 cm s-1. While
uncertainties are large, mainly because the particle decay
rates varied substantially between homes, the estimated Vd
value is reasonable. Vdis a strong function of particle size,
and there is considerable uncertainty regarding its value in
buildings(e.g.,for0.3-1µmdiameterparticles);experimental
values range up to 0.1 cm s-1, while theoretical and some
experimental values are as low as 0.001 cm s-1(15).
PredictedPMReductionsandCADR.PMreductionsare
calculated for house and room scenarios, AERs from 0.1 to
10h-1,andKpfrom0to0.001s-1(3.6h-1),assumingconstant
filterefficiency(ηp)75%).InFigure3A,thefilterisoperated
at a moderate speed, and air is assumed to be fully mixed
withinatypicallysized(2000squarefoot,186m2)U.S.house.
Results strongly depend on AER, Kp, and house or room
volume V. The line Kp) 0.0005 s-1corresponds to (average)
experimentalfindings.ThefilterreducesPMlevelsby>50%
onlyforAERse0.5andKpe0.0001s-1;onlysmallreductions
are obtained for AERs >1 h-1. In general, the filter is
undersizedinthisapplicationsincethetested(andsimulated)
unit is rated for a 510 ft2(47 m2) area. Also, the importance
of deposition processes alone should be noted (e.g., indoor
concentrations are only 27 and 23% of outdoor concentra-
tions for Kp) 0.0005 s-1, AER ) 0.5, and 1 h-1, respectively,
assuming no indoor sources and a penetration factor of 1).
Large values of Kpquickly remove PM relative to the filter’s
removal rate. In Figure 3B, the filter is placed in a room
largely sealed from the rest of the house (e.g., a bedroom
withthedoorclosed),butotherwiseconditionsarethesame.
Inthiscase,PMreductionsaremuchgreater,typicallyg75%,
for AER <1 h-1and most values of Kp. Again, high AERs tend
to defeat the filter’s performance.
Predictions of filter performance are not very sensitive to
variablesotherthanAER,house/roomsizeV,anddeposition
lossrateKp.Forexample,incomparisontothenominalcase
(Kp) 0.0005 s-1, AER ) 0.5 h-1), increasing filter efficiency
from75to100%increasesthePMreductionsbyanadditional
6 or 5% for the whole house or single room, respectively.
Increasing fan speed to a maximum (429 cfm ) 0.2 m3s-1)
increases removal by an additional 7 and 6% for the house
androom,respectively.Usingboth100%efficientfiltersand
the highest fan speed increases removal by an additional 14
and 9% for the house and room, respectively.
As shown elsewhere (12, 13, 15), filter performance
strongly depends on the AER and the volume cleaned. The
AER may need to be limited when using room filters; thus,
iftheventilationairisnotclean,naturalandforcedventilation
shouldbecontrolledbyclosingwindowsanddoors.Tomeet
comfort criteria in summer, an air conditioner that recir-
culates indoor air or that is equipped with a suitable filter
may be required.
In-use estimates of the clean air delivery rate (CADR)
depend on the volume represented by the concentration
measurements.Usingtheaveragehousesizetested(340m3),
PM0.3-1decaycurves,andeq2,theCADRis0.104m3s-1(220
cfm), about 30% lower than the filter’s rated capacity and
the measured air flow of the unit, although few participants
kept the unit on the highest speed. Still, this value indicates
little short-circuiting and good mixing. This CADR estimate
assumes that the houses were well-mixed, which should be
confirmed by additional information (e.g., concentration
FIGURE 2. Typical traces of PM0.3-1in late evening and early night
at house four measured without filter (top, 3/22/04) and with filter
(bottom, 3/25/04); 1 min observations.
TABLE 5. Estimated Half-Lives (h) of Particle Number
Concentrationsa
PM0.3-1
PM1-5
condition meanSDmeanSD
newwithout filter
with filter
without filter
with filter
without filter
with filter
0.73
0.27
0.41
0.33
0.57
0.30
0.47
0.06
0.41
0.24
0.42
0.15
0.63
1.65
0.68
1.25
0.66
1.45
0.34
1.03
0.10
0.37
0.23
0.67
used
all
aBased on three to six decay curves for each condition.

Page 7

measurementsinotherrooms).Fewstudieshavequantified
the spatial variation of PM within homes, although PM2.5
measurements in rooms in 10 homes in the UK showed a
high degree of spatial uniformity (40). Also, point measure-
ments using sampling periods exceeding ∼30 min appear
sufficiently long to represent concentrations within a room
(41). However, air flows may be complex (e.g., mixing may
be rapid within a room (especially when augmented by the
filter’s fan) but slower and variable between rooms).
Use, Operating, and Total Costs. Participants kept the
filter fans on continuously, as instructed, although they
adjustedspeedssomewhatoverthemonitoringperiod(e.g.,
site 1 kept speeds on high to turbo, site 2 was always on low,
site3wasmediumtohigh,andsite4wentfromlowtoturbo
and occasionally turned the unit off for periods of up to
severalhours).Participantstendedtoadjustspeedslesswith
time, presumably as they became used to the units.
Electricity consumption is estimated to cost $52 to $95
annually, depending on the fan speed and assuming con-
tinuous use and the prevailing electricity price in Detroit
($0.092kWh).TheHEPAfilterscost∼$90each,and(atleast)
annual replacement is advisable. Prefilters cost ∼$20 each,
andtheseshouldbereplacedorcleanedquarterly.Weassume
annual replacement of both filters. A central or room air
conditioner (AC) may be needed in summer to limit the air
exchangerates.IntheDetroitarea,electricitycostforasmall
window AC would be roughly $100. The initial costs of the
filter and air conditioner are $208 and $200, respectively.
Annualized costs, using an interest rate of 8% and assuming
asystemlifeof5yearsandthesecondhighestfanspeed,are
$236 for the filter and $151 for the air conditioner, or $386
for both. Operating costs represent the 74% of costs. Costs
excludeservicesofanelectricianorothersthatmaybeneeded
to install the AC. (The filter does not require installation
beyond plugging it in.) These costs apply for PM control in
a single room or section of a house (e.g., bedroom). Costs
would be higher to filter and cool a whole house.
Discussion
EmissionsfromPMsourcesinhomesmaycausemuchhigher
concentrationsthanoutdoorPMthatinfiltratesintothehome
(18, 42-45). Increases in coarse fraction PM concentrations
typically result from activities that generate and/or entrain
PM (e.g., movement of people, vacuuming, dusting, and
sweeping (44, 46-48)). Such activities, along with cooking,
increase PM2.5levels by roughly 3-6 µg m-3(45). Smoking
is the dominant contributor to indoor PM levels, increasing
PM2.5levelsby25-45µgm-3(43,45,49-51).Measurements
in Detroit homes have shown comparable impacts (52, 53).
Most tobacco smoke particles are between 0.1 and 0.36
µm in diameter (85% by mass), and only 1.5% exceeds 1 µm
(38); thus, tobacco impacts are largely limited to PM1. The
opticalcountsmadeinthehousesquantifiedonlythelarger
(g0.3)particlesassociatedwithtobaccosmoke,representing
∼15% by mass and ∼30% by number (37), and saturation of
the counter during smoking events led to significant un-
dercounting of PM number concentrations during these
events.CADRdeterminationsremainrobust,however,since
trendsofthelargerparticleslargelyreflectsmokegeneration
andremovalsbythefilter,andonlythechangesinPMlevels
(at levels within the useful range of the instrumentation)
were used to estimate first-order decay rates in eq 1. The
gravimetric measurements captured particles over a wide
size range and thus provide a more complete assessment of
PM levels; however, these time-integrated concentrations
do not allow identification of individual smoking events or
theuseoffirst-orderanalysesforestimatingdecayratesand
theCADR.Itwouldbeadvantageoustoutilizemoreadvanced
instrumentation in field studies that could measure the
smaller particles; however, optical and gravimetric measure-
ments were selected because they were field capable (e.g.,
small, quiet, capable of stable long-term unattended opera-
tion, etc.) and relatively modest in price.
In-use tests with and without filters in place effectively
showed the performance of the room air filters in smokers’
homes. The filters significantly diminished the duration of
concentrationpeaksassociatedwithsmokingevents.Results
were consistent in each house. Especially in cold/cool
weather, when AERs were lower as windows were closed,
filters provided significant reductions in gravimetric and
number concentrations of PM. Reductions were smaller in
warmweatherwhenwindowswereopenedandAERshigher.
While only approximate air exchange rates could be esti-
mated, results follow predictions using simple fully mixed
models (15). Preliminary CADR estimates derived from in-
use tests appeared reasonable, although concentration
measurements in other rooms are necessary to confirm
results. The filters were installed in the room with the most
important PM source, namely, smokers, and it is likely that
PMlevelswereloweredthroughoutthehouse.Filterairflow
decreased only slightly after 2-3 months of use, largely due
to loading on the prefilter, which could be easily cleaned
(washing or vacuuming) or replaced. To limit the number of
variables,onlyasingletypeoffilterunitwastested,butresults
probablyareapplicabletoothercommerciallyavailableroom
air filters units using similar filter types (coarse prefilter
followed by pleated paper HEPA filter).
Tracers for occupancy and ETS confirmed that smoking
frequency and ventilation did not change in the first trial
during cold weather, although smoking frequency indoors
and ventilation did change during warm weather. Measure-
mentorcontrolofthesevariablesisnecessaryintime-series
FIGURE 3. Predicted relative reductions in PM levels achieved by room air filter as a function of air exchange rate and Kp, the deposition
loss rate. Top line (Kp) 0) assumes no PM deposition. (A) Assumes 300 cfm (0.142 m3s-1) 75% efficient room air filter in 2000 SF (453
m3) house. (B) Same as panel A except that only one (224 SF, 51 m3) room is considered.

Page 8

comparisons, otherwise results may be invalidated by
changes in the number of cigarettes smoked, opening
windowsforadditionalventilation,andpossiblyotherfactors.
The percentage reductions achieved by room air filters
areunaffectedbyoutdoorPMlevelsorthestrengthofindoor
sources.However,indoorconcentrationswilldependonboth
factors. At low AERs, indoor concentrations may be limited
by the limited particle penetration to the indoors and
relativelyrapidlossesbydeposition.Ifoutdoorairispolluted,
then indoor concentrations may increase at high AERs if
indoorsourcesaresmall.InDetroit,ambientPM2.5andPM10
concentrationsaverageabout20and35µgm-3,respectively,
and peak levels can be several times higher. AERs below 1
h-1provide only marginal benefits from the standpoint of
PMreductions(Figure3).Ontheotherhand,concentrations
of gases, vapors, and moisture attributable to building
materials and occupant activities, not removed by the filter,
will increase as the AER is lowered. AERs should not be
excessively lowered without recognition of these issues.
The decay analysis for PM1-5was inconsistent because
fewwell-definedpeakswereobserved.Thisisnotsurprising
given that coarse fraction PM is generated by entrainment
and processes associated with human activities. The low
correlation between <1 and >1 µm diameter particles also
shows the role of different sources for these size fractions
(e.g., <1 µm PM trends related to smoking (and in some
cases,cooking,especiallybroiling,frying,andtoasting),while
the >1 µm particles likely resulted from entrainment and
resuspension from indoor activities).
We did not assess the effect of the filter’s fan speeds.
While participants were instructed to use the higher speeds,
participants used a range of speeds, including the lowest.
Most participants did keep the units on continuously.
Additional instructions, possibly a label placed on the unit,
and additional reinforcement and monitoring may help to
encouragetheuseofhigherfanspeeds.However,thebenefit
from higher fan speeds is predicted to be modest. Also, user
acceptance of the filters was very high. Several participants
indicated that they “loved” the filters, and several indicated
that they would buy units for themselves after the study
ended.
Overall, the tests demonstrated significant reductions in
PM in smoker’s homes, especially when air exchange rates
were limited. Filter performance was maintained over the
testing period, and user acceptance of the filters was high.
Such filters can reduce exposure of PM that has been
associated with the exacerbation of asthma and other
respiratory symptoms. Further research is necessary to
determine whether such filters improve respiratory health.
Acknowledgments
Wethanktheparticipantsfortheircooperation.MariaSalinas
and Kathy Edgren assisted recruitment and participant
orientation. Barbara Israel provided helpful suggestions.
SergeiChernyak,KevinFerrell,andYundaeYuhelpedinthe
lab. Gina Hatzivasilis provided editorial assistance. Portions
of the study were financially supported by the Michigan
Education and Research Center (funded by the National
Institute of Occupational Safety and Health, Grant T42
CC5410428), the American Chemistry Council (Grant 2401),
andtheMichiganCenterfortheEnvironmentandChildren’s
Health (funded by the National Institute of Environmental
Health Sciences and the U.S. Environmental Protection
Agency, Grant P01-ES09589).
Literature Cited
(1) Mannino, D.; Homa, D. M.; Akinbami L. J., et al. Surveillance
for asthmasUnited States, 1980-1999. MMWR Surveillance
Summaries 2002, 51, SS01, 1-13.
(2) NationalCenterforHealthStatistics.Asthmaprevalence,health
care use, and mortality, 2000-2001; Hyattsville, MD. Centers
for Disease Control and Prevention: Atlanta, GA, 2003.
(3) Baldwin,G.T.Assessingindoorasthmatriggersinurbanhomes
of children with asthma through a household walk-through.
Ph.D. Dissertation, The University of Michigan, Ann Arbor, MI,
2003.
(4) Hanley, J. T.; Ensor, D. S.; Smith, D. D.; Sparks, L. E. Fractional
aerosol filtration efficiency of in-duct ventilation air cleaners.
Indoor Air 1994, 4, 169-178.
(5) Foarde,K.K.;Hanley,J.T.;Ensor,D.S.;Roessler,P.Development
of a method for measuring single-pass bioaerosol removal
efficiencies of a room air cleaner. Aerosol Sci. Tech. 1999, 30,
223-234.
(6) Offermann, F. J.; Sextro, R. G.; Fisk, W. J.; Grimsrud, D. T.;
Nazaroff, W. W.; Nero, A. V.; Revzan, N. L.; Yater, J. Control of
respirable particles in indoor air with portable air cleaners.
Atmos. Environ. 1985, 19, 11, 1761-1771.
(7) American National Standards Institute/Association of Home
ApplianceManufacturers,“Americannationalstandardmethod
formeasuringperformanceofportablehouseholdelectriccord-
connected room air cleaners”, 1987, ANSI/AHAM AC-1-1988,
Chicago, Ill.
(8) Foarde, K. K.; Myers, E. A.; Hanley, J. T.; Enor, D. S.; Roessler,
P. F. Methodology to perform clean air delivery rate type
determinations with microbial aerosols. Aerosol Sci. Technol.
1999, 30, 235-245.
(9) Antonicelli,L.;Bilo,M.B.;Pucci,S.;Schou,C.;Bonifazi,F.Efficacy
of an air cleaning device equipped with a high efficiency
particulateairfilterinhousedustmiterespiratoryallergy.Allergy
1991, 46, 8, 594-600.
(10) Huang,S.W.Theeffectsofanaircleanerinthehomesofchildren
withperennialallergicrhinitis.Pediatr.AsthmaAllergyImmunol.
1993, 7, 2, 111-117.
(11) Shaughnessy, R. J.; Levetin, E.; Blocker, J.; Sublette, K. L.
Effectiveness of portable indoor air cleaners: sensory testing
results. Indoor Air 1994, 4, 170-188.
(12) Cheng,Y.S.;Lu,J.C.;Chen,T.R.Efficiencyofaportableindoor
air cleaner in removing pollens and fungal spores. Aerosol Sci.
Technol. 1998, 29, 2, 92-101.
(13) Green, R.; Simpson, A.; Cutovic, A.; Faragher, B.; Chapman, M.;
Woodcock,A.Theeffectofairfiltrationonairbornedogallergen.
Allergy 1999, 54, 5, 484-488.
(14) Emmerich, S. J.; Nabinger, S. J. Measurement and simulation
of the IAQ impact of particle air cleaners in a single-zone
building. International Journal of Heating, Ventilating, Air-
Conditioning and Refrigerating Research 2001, 7(3), 223-244.
(15) Fisk,W.J.;Faulkner,D.;Polonen,J.;Seppanen,O.Performance
andcostsofparticleairfiltrationtechnologies.IndoorAir2002,
12, 223-234.
(16) Offermann, F. J.; Louiselle, S. A.; Sextro, R. G. Performance of
air cleaners in residential forced air systems. ASHRAE J. 1992,
51-57.
(17) Jamriska,M.;Morawska,L.;Clark,B.A.Effectofventilationand
filtrationonsubmicrometerparticlesinanindoorenvironment.
Indoor Air 2000, 10, 19-26.
(18) Jamriska, M.; Morawska, L.; Ensor, D. S. Control strategies for
submicrometer particles indoors: model study of air filtration
and ventilation. Indoor Air 2003, 13, 96-105.
(19) Bowser, D.; Fugler, D.; Salares, V. The effects of improved
residential filtration on particulate exposure; Proceedings of
Indoor Air ‘99: Edinburgh, Scotland, August 8-13, 1999.
(20) Hanley, J. T.; Ensor, D. S.; Foarde, K. K.; Sparks, L. E. The effect
ofloadingdusttypeonthefiltrationefficiencyofelectrostatically
charged filters. Proceedings of Indoor Air ‘99: Edinburgh,
Scotland, August 8-13, 1999.
(21) Wood,R.A.;Flanagan,E.;VanNatta,M.;Chen,P.H.;Eggleston,
P. A. The effect of a HEPA room air cleaner on cat-induced
asthma and rhinitis [Abstract]. J. Allergy Clin. Immunol. 1997,
99, 1, S388.
(22) Brugge, D.; Vallarino, J.; Ascolillo, L.; Osgood, N. D.; Steinbach,
S.; Spengler, J. Comparison of multiple environmental factors
for asthmatic children in public housing. J. Indoor Air 2003, 13,
18-27.
(23) Pierce, W. M.; Janczewski, J. N.; Roethlisberger, B.; Pelton, M.;
Kunstel, K. Effectiveness of auxiliary air cleaners in reducing
ETS components in offices. ASHRAE J. 1996, 51-57.
(24) Reisman, R. E.; Mauriello, P. M.; Davis, G. E.; DeMasi, J. M. A
double-blind study of the effectiveness of a high-efficiency
particulate air (HEPA) filter in the treatment of patients with
perennialallergicrhinitisandasthma.J.AllergyClin.Immunol.
1990, 85, 1050-1057.

Page 9

(25) McDonald,E.;Cook,D.;Newman,T.;Griffith,L.;Cox,G.;Guyatt,
G.Effectofairfiltrationsystemsonasthma: Asystematicreview
of randomized trials. Chest 2002, 122, 5, 1535-1542.
(26) Warburton, C. J.; Niven, R. M.; Pickering, C. A.; Fletcher, A. M.;
Hepworth, J.; Francis, H. C. Domiciliary air filtration units,
symptoms,andlungfunctioninatopicasthmatics.Respir.Med.
1994, 88(10), 771-6.
(27) Reisman, R. E. Do air cleaners make a difference in treating
allergic disease in homes? Ann. Allergy Asthma Immun. 2001,
87, 6, 3, 41-43.
(28) Whirlpool, Whispure air purifiers, Features; http://www.whis-
pure.com/fea_home.html; accessed May 10, 2004.
(29) Anonymous.Aircleaners: behindthehype.ConsumerRep.2003,
October, 26-29.
(30) Peng, C. Y.; Batterman, S. Performance evaluation of a sorbent
tubesamplingmethodusingshortpaththermaldesorptionfor
volatileorganiccompounds.J.Environ.Monit.2000,2,4,313-
324.
(31) LaKind, J. S.; Jenkins, R. A.; Naiman, D. Q.; Ginevan, M. E.;
Graves,C.G.;Tardiff,R.G.Useofenvironmentaltobaccosmoke
constituents as markers for exposure. Risk Anal. 1999, 19, 3,
359-373.
(32) Daisey, J. M. Tracers for assessing exposure to environmental
tobaccosmoke: whataretheytracing?Environ.HealthPerspect.
1999, 107, 2, 319-327.
(33) Nelson, P. R.; Heavner, D. L.; Collie, B. B.; Maiolo, K. C.; Ogden,
M.W.Effectofventilationandsamplingtimeonenvironmental
tobacco smoke component ratios. Environ Sci. Technol. 1992,
26, 1909-1915.
(34) Rothberg,M.;Heloma,A.;Svinhufvud,J.;Ka ¨hko ¨nen,E.;Reijulaa,
K. Measurement and analysis of nicotine and other VOCs in
indoor air as an indicator of passive smoking.Ann. Occup. Hyg.
1998, 42, 2, 129-134.
(35) Gordon, S. M.; Wallace, L. A.; Brinkman, M. C.; Callahan, P. J.;
Kenny,D.V.Volatileorganiccompoundsasbreathbiomarkers
foractiveandpassivesmoking.Environ.HealthPerspect.2002,
110, 689-698.
(36) Riley, W. J.; McKone, T. E.; Lai, A. C. K.; Nazaroff, W. M. Indoor
particulate matter of outdoor origin: importance of size-
dependent removal mechanisms. Environ. Sci. Technol. 2002,
36, 200-207.
(37) Morawska, L.; Jamriska, M.; Bofinger N. D. Size characteristics
and aging of the environmental tobacco smoke. Sci. Total
Environ. 1997, 196, 43-55.
(38) Miller, S. L.; Nazaroff, W. W. Environmental tobacco smoke
particles in multizone indoor environments. Atmos. Environ.
2001, 35, 2053-2067.
(39) Howard-Reed,C.;Wallace,L.A.;Ott,W.R.Theeffectofopening
windowsonairchangeratesintwohomes.J.AirWasteManage.
Assoc. 2002, 52, 2, 147-159.
(40) Wigzell, E.; Kendall, M.; Nieuwenhuijsen, M. J. The spatial and
temporal variation of particulate matter within the home. J.
Exp. Environ. Epid. 2000, 10, 307-314.
(41) Klepeis, N. E. Validity of the uniform mixing assumption:
determininghumanexposuretoenvironmentaltobaccosmoke.
Environ. Health Perspect. 1999, 107, Suppl. 2, 357-363.
(42) Clayton, C. A.; Perritt, R. L.; Pelizzari, E. D.; Thomas, K. W.;
Whitmore, R. W.; Wallace, L. A.; Ozkaynak, H.; Spengler, J. D.
Particle total exposure assessment methodology (PTEAM)
study: distributions of aerosol and elemental concentrations
in personal, indoor, and outdoor air samples in a southern
Californiacommunity.J.Exp.Anal.Environ.Epid.1993,2,227-
250.
(43) Wallace, L. Indoor particles: a review. J. Air Waste Manage.
1996, 46, 98-126.
(44) Abt, E.; Such, H. H.; Allen, G.; Koutrakis, P. Characterization of
indoorparticlesources: Astudyconductedinthemetropolitan
Boston area. Environ. Health Perspect. 2000, 108, 1, 35-44.
(45) Wallace,L.Children’shealthsparticleconcentrationsininner-
city homes of children with asthma: the effect of smoking,
cooking,andoutdoorpollution.Environ.HealthPerspect.2003,
111, 1265-1272.
(46) Thatcher, T. Deposition, resuspenesion, and penetration of
particles within a residence. Atmos. Environ. 1995, 29, 1487-
1497.
(47) Vette, A. F.; Rea, A. W.; Lawless, P. A.; Rodes, C. E.; Evans, G.;
Highsmith, V. R.; Sheldon, L. Characterization of indoor-
outdoor aerosol concentration relationships during the Fresno
PM exposure studies. Aerosol Sci. Technol. 2000, 34, 1, 118-
126.
(48) Luoma, M.; Batterman, S. Characterization of particulate
emissions from occupant activities in offices. Indoor Air 2001,
11, 35-48.
(49) Dockery, D. W.; Spengler, J. D. Personal exposure to respirable
particulates and sulfates. J. Air Pollut. Control Assoc. 1981, 31,
2, 153-159.
(50) Ozkaynak, H. X.; Xue, J.; Spengler, J.; Wallace, L.; Pellizzari, E.;
Jenkins, P. Personal exposure to airborne particles and metals:
results from the Particle 6TEAM study in Riverside, CA. J. Exp.
Anal. Environ. Epid. 1996, 6, 57-78.
(51) Gauvin, S.; Reungoat, P.; Cassadou, S.; Dechenaux, J.; Momas,
I.; Just, J.; Zmirou, D. Contributions of indoor and outdoor
environmentstoPM2.5personalexposuresofchildrensVESTA
study. Sci. Total Environ. 2002, 297, 1-3, 175-81.
(52) Keeler,G.J.;Dvonch,T.;Yip,F.;Parker,E.A.;Israel,B.A.;Marsik,
F. J.; Morishita, M.; Barres, J. A.; Robins, T. G.; Brakefield-
Caldwell, W.; Sam, M. Assessment of personal exposures to
particulatematter(PM)amongchildrenwithasthmainDetroit,
MI as part of Community Action Against Asthma (CAAA).
Environ. Health Perspect. 2002, 110, 2, 173-181.
(53) Yip, F. Assessment of personal exposure to particulate matter
among asthmatic children in Detroit. Ph.D. Dissertation, The
University of Michigan, Ann Arbor, MI, 2004.
Receivedforreview July8,2004. Revisedmanuscriptreceived
February 3, 2005. Accepted June 17, 2005.
ES048951Q