United States
                  Environmental Protection
                  Agency
National Exposure
Research Laboratory
Research Triangle Park, NC 27rf t
                  Research and Development
EPA/600/SR-95/098  April 1997
4*EPA      Project Summary

                  The  Particle Team  (PTEAM)
                  Study: Analysis  of the  Data
                  H. Ozkaynak, J. Xue, R. Weker, D. Butler, P. Koutrakis, and J. Spengler
                    EPA and the California Air Resources
                   Board sponsored a study of human ex-
                   posure to inhalable particles in the Los
                   Angeles Basin. Results were reported
                   in Volumes I and II; this is a summary
                   of the third and final volume, dealing
                   with statistical analysis and physical
                   models. Nicotine and air exchange re-
                   sults are presented and analyzed.  A
                   model was developed to estimate the
                   penetration factors and decay rates for
                   inhalable (PM10) and fine (PM25) par-
                   ticles, for 15 elements associated with
                   each size fraction, and for polyaromatic
                   hydrocarbons (PAHs). The model was
                   also used to estimate source emission
                   rates for particles and elements pro-
                   duced by cigarettes and cooking, and
                   also for PAHs produced by cigarettes.
                     This Project Summary was developed
                   by EPA's National Exposure Research
                   Laboratory, Research Triangle Park, NC,
                   to  announce key findings of the re-
                   search project that is fully documented
                   in a separate report of the same title
                   (see Project Report ordering  informa-
                   tion at back).

                   Introduction
                     In 1986, Congress mandated that the
                   US EPA undertake a study of exposure to
                   particles. EPA's Atmospheric Research
                   and Exposure  Assessment Laboratory
                   (AREAL), now part of the National Expo-
 sure Research Laboratory (NERL), joined
 with California's Air Resources Board to
 sponsor a study in the Los Angeles Basin.
 The study was carried out primarily ,by the
 Research Triangle Institute (RTI) and the
 Harvard School of Public Health, with ad-
 ditional support from Lawrence Berkeley
 Laboratory  (LBL), Acurex, and AREAL.
 Small portable personal  monitors were
 designed to measure inhalable particles
 (aerodynamic diameter less than 10 pm,
 or PM10). In addition, stationary microenvi-
 ronmental  monitors were designed  to
 sample both PM,0 and PM25 (fine particles
 <2.5 fim in diameter). The personal and
 indoor samplers were equipped with  fil-
 ters to collect nicotine. Monitors were also
 developed to measure PAHs and  phtha-
 late esters. Air exchange rates in each
 home were measured using perfluorotracer
 (PFT) techniques. A total of 178 residents
 of Riverside, CA, took part in the study in
 the fall of 1990.
   The results of the study are presented
 in three volumes. Volume I (Pellizzari et
 al., 1993) provides a full description of  the
 procedures and presents summary popu-
 lation-weighted statistics for particles and
 elements. A Project Summary for Volume
 I  is also available (Pellizzari et al., 1993).
 Volume II (Sheldon et al., 1993) presents
 summary population-weighted statistics for
 PAHs and phthalate" esters.  Volume III
 presents summary statistics for  air  ex-
                                                                     Printed on Recycled Paper

-------
 change rates and nicotine concentrations,
 and also provides more detailed statistical
 analyses and physical models for all pa-
 rameters measured. This document is the
 Project Summary for Volume III.

 Procedure

 Measurement Methods
   A personal exposure monitor (PEM) was
 designed to collect PM10 using a sharp-cut
 Impactor with a circular set of holes 1.9
 mm In diameter. Particles are collected at
 a constant flow rate of 4 L/m on a 37-mm
 Teffon filter mounted below a greased im-
 pactor plate. The PEM consists of a soft
 canvas bag containing the pump and bat-
 tery pack that can be worn on the hip,
 stomach, lower back, or over the shoul-
 der. Nearly identical monitors were em-
 ployed for concurrent indoor and outdoor
 sampling of PM)0 and PMM. For the per-
 sonal and Indoor samples, a second filter
 treated with citric acid to collect nicotine
 was placed behind the first filter.
   The monitor for the PAHs and phthalate
 esters Included a glass cartridge contain-
 ing XAD-2  sorbent preceded by a quartz
 fiber filter.  A box containing  four Medo
 pumps sampled air at a constant flow rate
 of about 18 L/m during the 12-h monitor-
 Ing periods. Both the filter and the car-
 tridge were sonically extracted with meth-
 ylene  chloride. Analysis employed gas
 chromatography/mass spectrometry  (GC-
 MS) in the selected ion mode.
   Air exchange rates were measured us-
 ing perfluorotracer (PFT)  methods. Three
 continuously emitting sources of PFT in
 small boxes heated to 40°C were placed
 in each home 24 h before sampling be-
 gan. Volumes of the  homes  were  mea-
 sured at this time. At each  of the subse-
 quent visits (the beginning of each of the
 two 12-h measurement periods), collector
 tubes containing activated charcoal were
 placed in the  home  at three  sites.  Two
 tubes were in the main living area, one in
 the bedroom, and one near the center of
 the home.

 Study Design
  The main goal of the study was to esti-
 mate the frequency distribution of expo-
 sures to PM,. particles for all nonsmoking
 Riverside residents aged  ten and above.
A second major objective was to estimate
the frequency distribution of  concentra-
tions of PM 0 and PM2J!, PAHs, and phtha-
 late esters in residences and nearby out-
door air, e.g., back yards.  Other objec-
tives Included Identifying important indoor
sources and estimating  their emission
rates, determining the effect of outdoor air
on indoor concentrations, and estimating
 the  contribution of  personal activities to
 exposure.
   A three-stage probability sampling pro-
 cedure was adopted. In the first stage, 36
 areas within  Riverside were selected  for
 study with a probability proportional to
 population size in each area. Areas were
 characterized and stratified by income to
 ensure a wide socioeconomic representa-
 tion. In the second stage, an attempt was
 made to contact every household  within
 these areas and administer a short ques-
 tionnaire to determine eligibility for partici-
 pation and the frequency of certain  strati-
 fication variables such as employment and
 passive smoking. In the final  stage, re-
 spondents were selected for monitoring.
 Respondents  represented 139,000 ±
 16,000 (S.E.) nonsmoking Riverside resi-
 dents aged ten and  above.
   Each participant wore the PEM for two
 consecutive 12-h periods. (Actual  times
 monitored depended on participant  activi-
 ties and ranged from 8-14 h.) Concurrent
 PM,0 and PM25 samples were collected
 by the stationary indoor monitor (SIM) and
 stationary outdoor (ambient) monitor (SAM)
 at each home. This  resulted in 10 particle
 samples per household  (day  and  night
 samples from the  PEM10, SIM10, SIM25,
 SAM,0, and SAM2S). Air exchange rates
 were also calculated for each 12-h period.
 At a  subset of 125 homes (and  65 out-
 door areas near the  residences), monitors
 to measure indoor and outdoor PAHs and
 phthalate  esters were operated for each
 12-h period.
   Participants were asked to note activi-
 ties that might  involve increased particle
 levels (nearby smoking, cooking, garden-
 ing, etc.).  Following  each of the two 12-h
 monitoring periods, they answered an in-
 terviewer-administered recall questionnaire
 concerning their activities and locations
 during that time.
  Up to four participants per day could be
 monitored, requiring 48 days in the field.
 A central outdoor site was maintained over
 the entire period (Sept. 22-Nov. 9, 1990).
 The  site had two high-volume samplers
 (Wedding & Assoc.) with 10-um inlets (ac-
 tual outpoint about 9.0 urn); two dichoto-
 mous PM10 and PM25 samplers (Sierra-
 Andersen) (actual outpoints about 9.5 and
2.5 urn); one PEM and one SAM10 (actual
outpoint about 11.0  nm,  as measured in
 laboratory studies);  and  one SAM25 (ac-
tual outpoint = 2.5 urn).
  All  PEM, SIM, SAM, and dichotomous
sampler filters  (about 2500) were  ana-
 lyzed by energy-dispersive x-ray fluores-
cence (EDXRF) for a suite of 42 elements
 (Dzubay et a/.,  1988). The analysis was
carried out at EPA's AREAL in Research
 Triangle Park, NC. Some filters were ana-
 lyzed twice under blind conditions. A sub-
 set of about 100 filters was analyzed  by
 the LBL for quality assurance purposes.
 An additional set of about 600 citric-acid
 treated filters from personal and indoor
 samplers was analyzed for nicotine.
   All  filters  were weighed  onsite. Repli-
 cate weighings were required to be within
 ±4 ng/filter. Blank filters were  weighed,
 sent out with field samples, and reweighed
 along  with the field samples. Duplicate
 indoor and outdoor samples were collected
 at 10% of the homes. Duplicate SAM and
 PEM  samples were also collected at the
 central site. Duplicate PEM samples were
 also collected by EPA, RTI, and Harvard
 scientists while onsite.

 Results
   Of 632 permanent residences contacted,
 443 (70%) completed the screening inter-
 view.  Of these, 257 were asked to partici-
 pate and 178 (69%) agreed. More  than
 2750   particle samples were  collected,
 about 96% of those attempted.

 Quality of the Data
   Blank PEM and SIM/SAM filters (N = 51)
 showed consistent small increases in mass
 averaging 9.5 ± 8.4 ng; this value was
 subtracted from each field sample. Limits
 of detection (LODs), based on three times
 the standard deviation of the blanks, were
 on the order of  7 to  10 ng/m3.  All  field
 samples exceeded the LOD.
   Duplicate samples (N =  363)  showed
 excellent precision for all types of samplers
 at all  locations, with median relative stan-
 dard deviations ranging  from 2% to 4%.
   The collocated  samplers at the central
 site showed good agreement, with corre-
 lations between the three types  of sam-
 plers  ranging from 0.96 to  0.99. As had
 been  previously  noted, the central-site
 PEMs  collected  about  12% more mass
 than the dichotomous samplers, perhaps
 due to their higher outpoint  (11 \um com-
 pared to 9.5 urn) or to a particle "bounce"
 effect,  measured in the laboratory at less
 than 9%.
   Background levels of  elements on labo-
 ratory and field blanks were low. Analyses
 of standard reference materials (SRM 1832
 and 1833) were within  7% of the correct
 values for all 12 elements contained. Me-
 dian relative  standard  deviations (RSD)
for duplicates analyzed blindly by the prin-
cipal laboratory were better than 15% for
all 15 prevalent elements. Median RSDs
for duplicates analyzed by  the  principal
laboratory and by the  quality assurance
laboratory (LBL) were less than 21% for
all elements except manganese (76%) and

-------
copper (27%). The LBL reported 10% to
20% higher average values for 13 of 14
elements.
  For benzo-a-pyrene, the quantifiable limit
was 0.08 ng/m3, with a median precision
of 3%, a mean amount on the field blanks
of 0.0 ng, and a mean recovery of 95 ±
19%. Similar results were obtained for the
other PAHs and phthalate esters.

Concentrations
  Concentrations of particles and elements
have been reported (Clayton et al., 1993;
Ozkaynak et al., 1993;  Pellizzari et al.,
1993; Wallace ef al.,  1993). Population-
weighted daytime personal PMig concen-
trations averaged about 150 ng/m3, com-
pared to concurrent indoor and  outdoor
mean concentrations of  about 95 ng/m3.
The overnight personal  PM10  mean was
much lower (77  |Ag/m3) and more similar
to the indoor (63 (ig/m3) and outdoor (86
ng/m3) means. About 25% of the popula-
tion was  estimated to have exceeded the
24-h National Ambient Air  Quality Stan-
dard for PM10 of 150 ng/m3- Over 90% of
the population exceeded  the  California
Ambient  Air Quality Standard for PM10 of
50  ng/m3.
  Mean values of the fine (PM25) particle
mass were 48 and 49 ng/m3 for daytime
indoor and outdoor samples, and 36 and
50  ng/m3 for the overnight indoor and out-
door samples, respectively. Thus, fine par-
ticles accounted for about 50% of the total
PM10 mass both indoors and outdoors dur-
ing the day and about 60% both indoors
and outdoors at night.

Nicotine
  A total of 334 valid measurements were
obtained for the personal samples, and
230 for the indoor samples. About 30%
(176) of the 564 analyzed nicotine samples
exceeded the LOD of 0.15 jig/filter (corre-
sponding to a nominal value of about 0.05
ng/m3). Most of these were from personal
or  indoor samples associated with expo-
sure to cigarette smoke. Mean personal
and indoor  nicotine concentrations were
on  the order of 1 |ig/m3 for those samples
associated with  reported exposure to to-
bacco smoke but were below the LOD for
those samples with no reported exposure.
  A regression of indoor nicotine concen-
trations  on  the number  of cigarettes
smoked  in the home during the  monitor-
ing period  indicated that indoor nicotine
values increased by about 0.12 ng/m3 for
each cigarette reported smoked during the
monitoring  period. The ff value is 35.4%
(N = 227).
  A regression of personal nicotine levels
on minutes exposed to  cigarette smoke
suggested that personal nicotine expo-
sures increased by about 0.013 ng/m3 per
minute of reported exposure to cigarette
smoke. The Ff value is 36.6% (A/ = 334).

PAHs
  Median indoor and outdoor concentra-
tions ranged between 0.1 and 2 ng/m3 for
all but the two most volatile 3-ringed PAHs:
acenaphthylene (day and night medians
of 3.5 and 3.8 ng/m3 indoors and 1.8 and
6.9 ng/m3 outdoors) and phenanthrene (16
and 15 ng/m3 indoors and 8.8 and 12 ng/
m3 outdoors).  Little difference  was seen
between indoor  and outdoor concentra-
tions.

Phthalate Esters
   Median indoor values for four phthalate
esters ranged between 30 and 400 ng/m3;
they were below the detection limit for di-
n-octylphthalate. Median outdoor levels
were often below the detection limit, with
the highest value being 28 ng/m3 for di-2-
ethylhexylphthalate. Indoor levels of four
phthalates were about 2 to 15 times higher
than outdoor levels.

Air Exchange
   A total of 1010 12-h average air samples
were collected. There were 273 duplicate
pairs and 464 single observations.  Two
observations were outliers, resulting in 735
values  after averaging  the duplicates.
About 20%  of  the samples  had  PFT
amounts below the LOD.
   The 24-h average air exchange rates
were calculated for 175 Riverside homes
using the convention of assigning half the
LOD to values below the LOD. The geo-
metric mean of the air  exchange rates
was 0.97 tr1,  with a geometric standard
deviation (GSD)  of 2.18  (Figure 1).

Correlations
   The central site  appeared to  be a mod-
erately good estimator of outdoor particle
concentrations  throughout  the  city.
Spearman correlations of the central-site
concentrations measured by all three meth-
ods with outdoor  near-home  concentra-
tions as measured by the new samplers
ranged from 0.8 to 0.85 (p<0.00001). Lin-
ear regressions indicated that the central-
site readings could explain about 60% of
the variability observed  in the  near-home
outdoor concentrations (Figure 2).
   Outdoor concentrations  could explain
about 25%  to 30%  of the variability ob-
served in indoor concentrations (Figure 3).
Spearman correlations of near-home out-
door concentrations with indoor concen-
trations ranged from 0.5 to 0.6. Spearman
correlations of the central-site outdoor con-
centrations with indoor concentrations were
reduced somewhat (about 0.45 to 0.55).
  Outdoor concentrations  were  able  to
account for only about 16% of the vari-
ability in personal  exposures  (Figure 4).
This is understandable in view of the im-
portance of indoor activities such as smok-
ing, cooking, dusting,  and  vacuuming on
exposures to particles. The higher day-
time exposures were even less well rep-
resented by the outdoor concentrations,
whether measured near the home or  at
the central site.
  Indoor  concentrations accounted  for
about  half of the  variability in personal
exposures (Figure 5). However, neither
the indoor concentrations  alone,  nor the
outdoor concentrations alone, nor time-
weighted averages of  indoor and  outdoor
concentrations  could  do  more than ex-
plain about two-thirds of the observed vari-
ability  in personal  exposures. It appears
that the remaining portion of personal ex-
posure arises from personal activities  or
unmeasured microenvironments that are
not well represented  by fixed indoor  or
outdoor monitors.
  One of the variables most highly corre-
lated with particle levels in the home was
an'estimated "dirt level." The technicians'
estimate of dirt level was made while visit-
ing each  house. Two  technicians carried
out all the measurements. They estimated
dirt and dust levels  on a 7-point scale,
and "calibrated" themselves by experiment-
ing on several  Boston homes before go-
ing to  Riverside. The 24-h averages  of
personal  and  indoor  particles and also
indoor nicotine were significantly associ-
ated with estimated dirt level.
  The  mean  indoor  concentrations  of
PAHs  in  homes with  smoking appeared
generally higher than in homes without
smoking. Student t-tests showed that  for
10 out  of 12 PAHs (and  for  di-n-
butylphthalate), the difference between the
geometric mean concentrations in smok-
ing vs. nonsmoking  homes was statisti-
cally significant, usually at  p < 0.0001.

Mass-Balance Model
  A model developed in Koutrakis et al.
(1992) was solved using  nonlinear least
squares to estimate  penetration factors,
decay rates, and source strengths for par-
ticles and elements from  both size frac-
tions. In this model, which assumes per-

-------
      20
      16
      12
                          I
                                      I
                                         I
           -2.1          -1.1         -0.1           0.9

                       Log 24-h Average Air Exchange Rate (1/h)
                                                     1.9
                                                                  2.9
          Figure 1. Normal fit to the logarithms of the overnight air exchange rates, assigning
                  values ofLOD/2 to all nondetectedsamples. The geometric mean = 0.97 h''.
feet Instantaneous mixing and steady-state
conditions throughout each 12-h monitor-
Ing period, the indoor concentration of par-
ticles or elements is given by
                                                                             /Vclg   = number of cigarettes smoked
                                                                                     during monitoring period

                                                                             Sclg   = mass of elements or particles
                                                                                     generated per cigarette
                                                                                     smoked (ng/cig or ng/cig)

                                                                             TOX*   = time spent cooking (min) dur-
                                                                                     ing monitoring period

                                                                             Scools   = mass of elements or particles
                                                                                     generated per minute of cook-
                                                                                     ing (ng/min  or ng/min)

                                                                             Bother  = mass flux of elements or par-
                                                                                     ticles from all other indoor
                                                                                     sources (ng/h or ng/h)

                                                                            With these changes, the equation for
                                                                          the indoor concentration due to these in-
                                                                          door sources becomes
                                                                                 C. = •
   -fa
where
   P
   a
              a+k
                                 [1]
i Indoor concentration (ng/m3
 for elements, ng/m3 for par-
 ticles)
r penetration coefficient
; air exchange rate (h-1)
i outdoor concentration (ng/m3
 or ng/m3)
   Qfe    = mass flux generated by in-
           door sources (ng/h or ng/h)
   V    = volume of room or house (m3)
   k     = decay rate due to diffusion or
           sedimentation (h-1)

  From initial multivariate analyses, the
most important indoor sources appeared
to be smoking and cooking. Therefore the
indoor  source term Qte was replaced  by
the following expression:
where
   t
                                                     duration of the monitoring pe-
                                                     riod (h)
                                                                                       a + k
                                                                                        (a + K)Vt
                                                                       (a + k)V

                                                                          [2]
  The indoor and outdoor concentrations,
number of cigarettes smoked,  monitoring
duration, time spent cooking, house vol-
umes, and air exchange  rates were all
measured or  recorded. The penetration
factor, decay rates, and source strengths
for smoking, cooking, and all other indoor
sources  (Qott]  ) were estimated using a
nonlinear model (NLIN in SAS software).
The  Gauss-Newton  approximation  tech-
nique was chosen to regress the residu-
als onto the partial derivatives of the model
with  respect to the unknown parameters
until  the estimates converge. On the first
run,  the penetration  coefficients were al-
lowed to "float" (no requirement was made
that  they be < 1).  Since nearly all coeffi-
cients came out close to one,  a  second
run was made bounding them from above
by one. The NLIN program provides sta-
tistical uncertainties (upper and lower 95%
confidence intervals) for all parameter es-
timates. However,  it should be noted that
these uncertainties assume perfect  mea-
surements and are  therefore  underesti-
mates of the true uncertainties.
  Results are presented in Table 1 for the
combined day and night samples.  Pen-
etration factors are very close to unity for
nearly all particles and elements. The cal-
culated  decay rate  for fine particles  is
0.39 ±  0.16 h-1, and for PM10  is  0.65 ±
0.28 rr1. Each cigarette emits 22 ± 8 mg
of PM10  on average, about two-thirds of
which (14 + 4 mg) is in the fine fraction.
Cooking emits 4.1 ±1.6 mg/min of

-------
 I
 I
 •p
 1
       600
       500
       400
300
       200
       700
                                                           Backyard = 1.03'Central + 17.6
                                                           R2=0.57 n = 323
                             50
                                          100
150
                                       Central site reference monitor mean (\\.g/m )
Figure 2. Central-site mean of two dichotomous samplers vs. residential outdoor monitors.
         Ft1 = 57%
                                                                                       200
                                                                                                           250
  -c
  C\l
      600
      500
      400
      300
      200
      700
                                                          Indoor = 0.54 *Outdoor + 32
                                                            z= 27% (n=309)
                         100
                                   200            300             400

                                  Average 12-h outdoor concentration (}ig/m3)
Figure 3. Indoor vs. outdoor PMW concentrations.
         Ft2 =27%
                                                                                          500
                                                                                                          600

-------
       500
        400
       300
       200
        roo
                                                                   PERS = 0.54 "OUT'+ 62
                            100
                                            200             300
                                            Backyard concentrations (iig/m3)
                                                                             400
500
600
FJgun 4. Personal exposures vs. residential (backyard) outdoor PM,0 concentrations.
        R*- 16%
       500
        400
       300
       200
        roo
                                                         PEPS = 0.91* Indoor + 39
                                                              49% (n=321)
                            700             200              300             400
                                         Average 12-h indoor concentration (\ig/m3)
FJguraS. Personal exposures vs. residential indoor PM,aconcentrations.
                                                                                             500
                                                                                                             600

-------
     *
     §,
     CO1
     !
     co
••5

Uj
€   &
!   «
«   I
§   s
1
s

          $
fi
                 NC cS ci CM
                 i~ CM
                 oo o CM o>
                  co *- CM CM
                  CM co ci cS
                  O M"f- O
                  i-: t-^cici
                   - Q «5 to
                                         cQidtoioio   10   co«o         CM
                                           c\j   t—      Y-      co
                                    v- o -q- 
                                 CMOcot35M;co'<»O5CMcq   N. '

                                          ! CM »-        ^      CM
                                                                                    M; CS O N. O O CM


                                                                                    CM oo   c\j      co1
                                                                  cq ••-; cq CM cq «q o>


                                                                  CM CM      *~
                                 Oo>--tqcoco''*O5iq>-   tqtO'r-OcqcqCjC)    05 cq       to c> -^ CM •<* id cd •<-: ci
                                                --
                  QK.CV1CO
                  c> co c5 c>
                                                                CM
                                                                             T-IO    tqcqiqiqO5M;M;


                                                                                     •q- CM >-        CM
                                 -^ ci ci CM co -r-: c> cjj
                                            cqC5CMT~tqtqcocq    »->-iqtqM;'^;^;
                                                            'ci    tdN^idcdcicooi
                                                                   •^- CM -7   •—   M-
                                 CO-*^COCO'»^COOiOlOCO    tOlOCM^~C5COC\|O5    COtOlOtOt^
                                 cocicirvlscpcd^NCgi    ids^-C'^^co'ci    COOIO^CM^^

                                                                          ^         CM ••-           "•
                                         «i co CM
                                                         CM
                                                                -^-
                                                               i CM o co CM cn co 05
                                                                   ^§^t^S?
ci Cji ci cS     ci ci ci ci ci Q ci cS cS ci   ci ci ci ci ci ci ci ci   cS ci ci ci ci ci ci
                                                                                     C\I CO t
                                                                                     to to c
                                                                       > to^CMCOCO
cicicsd S.cot-:cicicScicicicicic>   cicicicicicicic! gjciciciciciciCM
                             11
                  ci cS ci cj
                                                                                     *- co o co to t~ •*
                                                                                     cq to cq cq 05 cq ^
                                  ci ci ci ci ci ci ci ci ci ci   c5 ci ci ci ci ci ci ci   ci ci ci c! ci ci ci
                                                              88888888   8888885
                                                     . O O   ^— T— Y^-Y^t— t— Of-   T— '

-------
 Inhalable particles, of which about 40%
 (1.7 ± 0.6 mg/min) is in the fine fraction.
 All elements emitted by cooking were lim-
 ited almost completely to the coarse frac-
 tion. Sources  other than cooking  and
 smoking emit about 5.6 ± 3.1  mg/h of
 PM , of which only about 1.1 mg/h ± 1.0
 (20%) is In the fine fraction.
   Decay rates for elements associated
 with the fine fraction were generally lower
 than for elements associated with the
 coarse fraction, as would be expected.
 For example, sulfur, which has the lowest
 mass median diameter of all the elements,
 had calculated decay rates of 0.16 + 0.04
 and 0.21 ± 0.04 h-' for the PM,^ and PM10
 fractions, respectively. The crustal elements
 (Ca, Al, Mn, Fe), on the  other hand, had
 decay rates ranging from 0.6 to 0.8 fr1.
   A similar model was developed for the
 PAHs and phthalate esters. However, the
 model differs in not including a cooking
 source term because initial regression
 analyses did not identify cooking as an
 important source of PAHs. Also, because
 of a much lower sample size for the PAH
 and phthalate  measurements  (only 60
 homes with outdoor measurements, com-
 pared to 160 for the particles), the param-
 eter estimates show much larger uncer-
 tainties.
   Results showed that penetration factors
 for most PAHs were  very close to  one
 (Table 2). Because of the very large un-
 certainties involved in trying to estimate
 four unknown parameters (penetration co-
 efficients, decay  rates,  smoking  source
 strengths, and  flux from other indoor
 sources), the penetration coefficients were
 all  set  equal to one and the remaining
 three parameters in Table 2 were calcu-
 lated using the nonlinear algorithm.
  The estimated average decay rates for
 PAHs ranged from 0.4 to 1.6 h-1 with size-
 able variation. No apparent dependence
 on  volatility  was noted. Estimated decay
 rates for the phthalates  had very large
 uncertainties.
  Estimated indoor source strengths for
 smoking forbenzo(a)anthracene, chrysene,
 benzo(e)pyrene,benzo(a)pyrene,
 benzo(ghi)perylene, and  coronene were
 122, 192, 86, 264, 244, and 245 ng/ciga-
 rette, respectively. Smoking contributed 20%
 to 40%  of the total concentrations of eight
 PAHs in homes reporting smoking. Consid-
 ering all homes,  outdoor  air contributed
 more than half the total concentrations of
 six PAHs, mostly the  less  volatile ones,
and •tother"  (unidentified)  indoor sources
contributed more than half of three volatile
 PAHs.
  These other  Indoor sources were re-
sponsible for 97%  to 99%  of  the  total
concentrations measured for diethyl  ph-
thalate and di-n-butyl phthalate. Smoking
was not indicated as an indoor source for
phthalates.
   Other (unidentified) indoor sources were
found to be very important for both of the
phthalates and for the volatile  PAHs, but
the  estimates of source  strengths were
highly uncertain.
   The physical  models had reasonably
good fit to particulate PAHs, and the cor-
relation between predicted and observed
concentrations averaged about 0.7.
   Outdoor air was the major source of
indoor particles,  providing about three-
fourths of fine particles and  two-thirds of
inhalable particles in the average home. It
was also  the major source for most ele-
ments, providing 70% to 100% of the ob-
served indoor concentrations for 12 of the
15 elements. Only copper and chlorine
were predominantly due to indoor sources
in both the fine particle and inhalable par-
ticle fractions.
   Unidentified indoor sources  accounted
for most  of the  remaining  particle and
elemental mass  collected on the indoor
monitors.  The nature of these  sources is
not yet understood. They do not include
smoking, other combustion sources, cook-
ing,  dusting, vacuuming,  spraying,  or
cleaning, since all these sources together
account  for  less  than the  unidentified
sources.   For example, the unidentified
sources accounted for  26%  of the aver-
age  indoor PM  particles, whereas smok-
ing accounted for 4% and cooking for 5%.
   Of the identified indoor sources, the two
most important were smoking  and cook-
ing (Figures 6 and 7). Smoking was esti-
mated  to  increase 12-h  average  indoor
concentrations of  PM10 and PM26 by 2
and  1.5 pg/m3 per cigarette,  respectively.
Homes with smokers averaged about 30
p.g/m3 higher  levels of PM10  than  homes
without smokers. Most of this increase
was  in the fine fraction. Cooking increased
indoor concentrations of PM10 by about 20
pg/m3,  with most of the increase in  the
coarse particles.
   Emission profiles for elements were ob-
tained for  smoking and for cooking. Major
elements  emitted by  cigarettes were  po-
tassium, chlorine, and calcium. Elements
associated with cooking included  alumi-
num, iron, calcium, and  chlorine.
  Other household activities such as vacu-
uming  and dusting appeared to  make
smaller contributions to indoor particle lev-
els. An interesting finding was  that com-
muting and working outside the home re-
sulted in tower particle exposures than for
persons staying at home.
  As with the particle mass, daytime per-
sonal exposures  to 14 of 15 elements
were consistently higher than either in-
door or outdoor concentrations. At night,
levels of the elements were similar in all
three types of samples.

Discussion

Source of Excess Personal
Exposure
  The more than 50%  increase in  day-
time personal exposures compared to con-
current  indoor or outdoor  concentrations
suggested  that personal activities  were
important determinants of exposure. How-
ever, the nature of this "personal cloud" of
particles has not yet been  determined.
Scanning electron microscopy was under-
taken on 138 personal filters.  Skin flakes
were common on many filters. A prelimi-
nary analysis suggested that the average
number of  skin  flakes per filter  was
120,000 to 150,000. The mass of some
personal filters may have been consider-
ably increased by unusually  large  num-
bers of  skin flakes. However,  attempts to
calculate the mass of skin  flakes  from
estimates of their volume and density sug-
gest an average contribution to the mass
of only  about 4 pg/m3,  less than 10% of
the mass of the average personal cloud.
  Another approach to the composition of
the personal cloud is elemental analysis,
using x-ray fluorescence. Analysis of all
personal and indoor filters showed that 14
of 15 elements were elevated by values
of 50% to  100% in the personal filters
compared to the indoor filters (Figure 8).
This observation suggests that a compo-
nent of the personal cloud is an aerosol of
the same general composition as the in-
door aerosol. This could be particles cre-
ated  by activities,  e.g., cooking, or
reentrained household  dust from motion
(walking across carpets, sitting on uphol-
stered furniture). House dust is a mixture
of airborne outdoor aerosols, tracked-in
soil and road dust, and aerosols produced
by indoor sources. As such, it should  con-
tain crustal elements from soil, lead and
bromine from automobiles, and other ele-
ments  from combustion sources.   This
would be consistent with the observation
that nearly all elements were  elevated in
personal samples. The  fact that personal
overnight samples showed smaller mass
increases than  the personal  daytime
samples is also  consistent with the fact
that  the participants were sleeping for
much of the 12-h overnight monitoring pe-
riod, and were thus not engaging in these
particle-generating or reentraining activi-
ties.

-------
to
a
•3

I
1
£
5
I
£
co
CO
 fe

1
•o
 ci
 O)
 &
Uj

I
 CO
 CM
 Q>

 s
        .S1
        t
          :>
        CO
            -CO


            w


            TO1



            Q
            €
            I

            1

            I
                       N. >-
                       TJ-^J-
                       CM CM
                       CM co
                       co CM

                       $CM"
                                  C\|
                                                 O>
                                                 CO
                               COCOCMCON*COtO'*—OCM**—
                               ^J-CMCO^*-^N.COCM
                               CO Vl                     *Q Cft


                                                         CM CM
                                                          CM
                                                          ~
                       •g O
                                CO t- CM CM CO "J CO 3- IO O O
                                TJ- co CM 01 oo c6 •* S- 5- ^- *-
                                CM ••-•••- i-   CM ^- CM CM
                       IOCO    OCM^t-O
                       CM^    "3-CMCQOO

                       cS cS    ci cS ci c5 cS c> c> ca ci c5 cs
                                                            -co
C\l T—    Oi^J-COCOCQt—CO*-fSi-'iOCJ3
COCO    •^•COCMlO^I-CO'»-U>^J-CMO)

c> cS    cJ cJ cJ d c> cj cJ cS d c> c\j
                        O O   O O i
                                                          c>o
                                                                              1

-------
                          Cooking 3%

                                Other Indoor 7%
                                                                                                    Other Indoor 8%
                                                                                                 Smoking 5%
                         Cooking 3%
                                  Other Indoor 16%
                                                                                                    Other Indoor 16%
                                                                                            Smoking 4%
N - 81 Samples from 31 homes

Figure 6.  Sources of fine particles (top) and Inhalable particles
          (bottom) In 31 homes with smoking. Relative uncertainties
          In the estimates can be determined from the upper and
          tower confidence limits for the source strengths provided in
          Tabla 1.
                          N = 62 Samples from 33 homes

                        Figure 7.  Sources of fine particles (top) and inhalable particles
                                (bottom) in homes with cooking. Relative uncertainties in
                                the estimates can be determined from the upper and lower
                                confidence limits for the source strengths provided in Table 1.
  The measurements at the central site
showed good agreement with the outdoor
measurements at  homes  throughout the
City of Riverside, indicating that a single
central-site PM10 monitor can characterize
a large urban area adequately. Although
the correlations of indoor air concentra-
tions with outdoor air are lower,  there is
evidence that outdoor air PM10 concentra-
tions can affect indoor air concentrations.
   The nonlinear least squares method of
solving the mass-balance model improved
on previous formulations in making fewer
arbitrary assumptions and solving for all
unknown parameters simultaneously. An
Interesting result from this effort was the
finding that the penetration factor was very
close to  one for nearly all particles, ele-
ments, PAHs and phthalate esters.

Conclusions
  The personal and  microenvironmental
monitors designed especially for this study
performed well. About 96% of all samples
attempted were collected and median pre-
cision was 2% to 4%.
  The major finding of the study was the
50% increase in daytime  personal expo-
sures to PM10  compared to indoor and
outdoor concentrations. The increase ap-
pears to be due to personal activities such
as dusting, vacuuming, cooking, and shar-
ing a home with a smoker. This suggests
that reduction of dust levels in the home
could decrease exposure to airborne par-
ticles.
  A mass-balance  model provided  esti-
mates  for the source strengths of ciga-
rettes and cooking  for particles and ele-
ments  in two size  fractions.  The  model
also provided estimates for  the  source
strengths of  cigarettes  for a number of
PAHs.
                                                            10

-------
  Elements in the Personal Cloud
                    20
                                                                 80
                                   40             60

                                 Percent increase in personal cloud


Figure 8. Increased concentrations of elements in the personal vs. the indoor samples.
100
                                                                                               120
References
  Clayton, C.A., Perritt, R.L.,  Pellizzari,
    E.D., Thomas, K.W., Whitmore, R.W.,
    Ozkaynak,  H.,  Spengler,  J.D.,  and
    Wallace, L.A. Particle Total Exposure
    Assessment Methodology  (PTEAM)
    Study:  Distributions of Aerosol  and
    Elemental Concentrations in Personal,
    Indoor, and Outdoor Air Samples in a
    Southern California Community. J Ex-
    posure Analysis and  Environmental
    Epidemiology 3:227-250, 1993.
  Dzubay,  T.,  Stevens, R.K., Gordon,
    G.E., Olmez, I., Sheffield,  A.E.,  and
    Courteney,  S.W.I. "A Composite Re-
    ceptor Model Applied to Philadelphia
    Aerosol," Environ. Sci. & Tech. 22:46-
    52, 1988.
  Koutrakis,  P., Briggs,  S.L.K.,  and
    Leaderer, B.P. "Source Apportionment
    of Indoor  Aerosols in Suffolk  and
    Onondaga  Counties, New York"
    Environ. Sci. Tech. 26:521-27. 1992.
                                           Ozkaynak, H., Spengler, J.D., Xue, J.,
                                             Koutrakis,  P.,  Pellizzari,  E.D.,  and
                                             Wallace, L.A. Sources  and Factors
                                             Influencing Personal and Indoor Ex-
                                             posures  to  Particles: Findings  from
                                             the Particle TEAM Pilot Study. In: In-
                                             door Air '93: Proceedings of the 6th
                                             International Conference on Indoor Air
                                             Quality and Climate. Vol. 3: pp. 457-
                                             462, 1993.
                                           Pellizzari, E.D., Thomas, K.W., Clayton,
                                             C.A., Whitmore, R.W., Shores, R.C.,
                                             Zelon,  H.S., and Perritt, R.L. Particle
                                             Total Exposure Assessment Method-
                                             ology (PTEAM): Riverside, California
                                             Pilot Study.  Vol. I. Project Summary.
                                             EPA/600/SR-93/050.  Research  Tri-
                                             angle Park, NC, 1993. Report EPA/
                                             600/R-93/050 NTIS # PB 93-166957.
                                             NTIS. Springfield VA, 1993.
     Sheldon,  L, Clayton, A.,  Keever, J.,
       Perritt, R., and Whitaker, D. PTEAM:
       Monitoring of Phthalates and PAHs in
       Indoor and Outdoor Air Samples in
       Riverside, California.  Vol. II. Air Re-
       sources  Board, Research  Division.
       Sacramento, CA, 1993.
     Wallace, L., Ozkaynak,  H.,  Spengler,
       J.D., Pellizzari, E.D., and Jenkins, P.
       Indoor, Outdoor, and Personal Air Ex-
       posures  to Particles,  Elements, and
       Nicotine for 178 Residents of River-
       side, California. In: Indoor Air '93: Pro-
       ceedings of the 6th International Con-
       ference on Indoor Air Quality and Cli-
       mate. Vol. 3: pp. 445-450. Helsinki
       University of Technology. Espoo, Fin-
       land, 1993.
                                                           11

-------
   H, Ozkaynak, J. Xue, Ft. Weker, D. Butler, P. Koutrakis, and J. Spengler are with
     Harvard University School of Public Health, Boston, MA 02115.
   Andrew Undstrom is the EPA Project Officer (see below).
   The complete report, entitled "The Particle Team (PTEAM) Study: Analysis of the
     Data—Final Report, Volume III," (Order No. PB97-102 495; Cost: $85.00, subject
     to change) will be available only from:
           National Technical Information Service
           5285 Port Royal Road
           Springfield, VA 22161
           Telephone: 703-487-4650
   The EPA Project Officer can be contacted at:
           National Exposure Research Laboratory
           U.S. Environmental Protection Agency
           Research Triangle Park, NC 27711
United States
Environmental Protection Agency
Center for Environmental Research Information
Cincinnati, OH 45268

Official Business
Penalty for Private Use $300
     BULK RATE
POSTAGE & FEES PAID
         EPA
   PERMIT NO. G-35
EPA/6QO/SR-95/098

-------