United States
               Environmental Protection
               Agency
Environmental Monitoring
Systems Laboratory
Las Vegas, NV 89193-3478
               Research and Development
EPA/600/SR-92/094  September 1992
EPA      Project  Summary
               Tests  of Indoor
               Air  Quality  Sinks
               James Quackenboss, Janet Remmers, James McHugh and Karin Bauer
                 Experiments were  conducted in  a
               room-size test chamber to determine
               the sink  effects of selected materials
               on  indoor air concentrations of p-
               dichlorobenzene (PDCB). These effects
               might alter pollutant behavior from that
               predicted using simple indoor air qual-
               ity models, by reducing the peak con-
               centrations during  source usage and
               by  increasing the ventilation rates
               needed to reduce pollutant concentra-
               tions. Four experimental conditions
               were tested:  empty chamber (painted
               gypsum wallboard), carpeting only, car-
               peting and drapes,  and carpeting plus
               a full-size bed with a comforter. Cham-
               ber temperature and relative humidity
               were controlled; atmospheric pressure
               was monitored. Air exchange rates were
               monitored using the tracer-gas decay
               method.  Chamber air, sampled at six
               minute intervals, was analyzed using a
               gas chromatograph. Sink effects were
               estimated by comparing decay curves
               fit to the tracer gas (SF6) and the PDCB
               concentrations. The influence  of sink
               effects on indoor air concentrations was
               illustrated by estimating the ventilation
               rates required to reduce PDCB concen-
               trations from 14 ppm  to 2 ppm. In the
               absence of sink effects, the theoretical
               rate was  1.95 air  changes  per hour
               (ACH). With sinks, the rates predicted
               for  each  experimental condition were:
               2.5  (empty test chamber),  3.21 (carpet
               only), 6.64 (carpet and  drapes), and 3.75
               ACH (carpet and bed).
                 This Project Summary was developed
               by  EPA's  Environmental Monitoring
 Systems Laboratory, Las Vegas, NV, to
 announce key findings of the research
 project that is fully documented in  a
 separate  report of the same title (see
 Project Report ordering information at
 back).

 Introduction
   The Toxic Substances Control  Act
 (TSCA) mandates the assessment of risks
 associated with the manufacture and use
 of new and existing chemicals. In 1985,
 the Interagency Committee on Indoor Air
 Quality called for  developing  an under-
 standing  of human exposures to indoor
 air pollutants, the contributions of various
 energy conservation  measures, and the
 impact of introducing new consumer prod-
 ucts and building materials. This effort re-
 quires assessing exposure to chemical re-
 leases from consumer products and build-
 ing materials in indoor environments un-
 der various conditions.
   Due to the number and variety of prod-
 ucts on the market, it would be extremely
 costly to rely exclusively on the monitoring
 of indoor  air concentrations  in order to
 determine all possible exposure implica-
 tions of these products. A preferable op-
 tion is the use of predictive indoor air
 quality models to estimate the air concen-
 trations or. personal  exposures that  can
 be expected under various conditions.
   One of  the most commonly used  pre-
 dictive mathematical models is the simple
 dilution model. In  a well-mixed chamber
 with a constant air exchange rate (A), the
 air concentration of a nonreactive pollut-
 ant  in air (C(), generated at a constant
                                                                  Printed on Recycled Paper

-------
emission  rate (G), can be predicted at
time  (t)  by the following  mathematical
model1:
        C, - G(1 - e-*1) / (AV)
(1)
  where
      C,  -  air concentration at time (t),
      G  «  emission rate,
      A  -  air exchange rate,
      V  -  exchange volume, and
       t  -  time.

  After the source is removed,  the  air
concentration decays exponentially and is
defined by the following equation:
               ff*
(2)
where C0 is the initial air concentration at
t-0, the time at which the source is re-
moved.
  The dilution mode! assumes that no pol-
lutant is tost within the chamber from any
mechanisms other than  dilution (e.g.,
through chemical decomposition, chemi-
cal reaction, or mass transfer between the
gas or vapor  phase and  solid surfaces
inside the chamber). These other mecha-
nisms are often  referred  to  as "sinks."
The dilution model also assumes that the
inlet air stream delivers  "clean" air to the
chamber at the same temperature and
pressure as those  of the outlet stream,
and that there is complete mixing of the
chamber air.
  The application of the dilution model to
predict exposure to a pollutant in an in-
door environment is complex if materials
(such as ftoor and wall coverings, window
drapes, and upholstered furniture) appre-
ciably adsorb or absorb the pollutant. For
purposes of this study, a  "sink effect" is
defined  as the mass transfer of an air
pollutant between its gas or vapor phase
and any surfaces inside the chamber. The
adsorption of the nonreactive pollutant to,
and  subsequent desorption  from,
nonreactive surfaces is assumed to be a
reversible process.
  The objectives of this project were two-
fold. First, experimental tests in a environ-
mental chamber were performed to as-
sess the sink effects of selected surfaces
and building materials on  the concentra-
tions  of an indoor air pollutant. Second,
the importance of these sink effects to
existing predictive mathematical models
of indoor air pollutants was to be deter-
mined. The project report presents  the
data for 12 experimental runs using one
chemical, p-dichlorobenzene (PDCB). Re-
sulls of 11  of these runs are presented in
the body of the report. Due to problems in
1  Ootm.J.E. 1987. Models and Statistical Methods for
  Gaseous Emission Testing of Finite Sources in Well-
  mixed Chambers. Atmos. Environ. 21:425-430.
run 11, all data and results pertaining to
that run are considered suspect and are
presented  separately (in Appendix  H of
the project report).
   PDCB was selected for study because
it is widely used in the indoor environment
as an insecticide (moth repellant), a disin-
fectant, or a deodorant (room freshener).
It was also selected to minimize the sink
effects from chemical reaction or degra-
dation mechanisms, allowing the study to
focus on adsorption and desorption mecha-
nisms.

Procedure
   Experiments were conducted in a room-
size (1,261-ft3)  test chamber at the Mid-
west Research  Institute's (MRI's) Air Con-
sumer Exposure (ACE) Laboratory under
stable environmental conditions. The test
chamber simulates  a residential room
where household  consumer  products
would be used.  Interior surfaces were cov-
ered  with  gypsum  wallboard.  The wall-
board was sealed with ready-mix joint com-
pound and painted with  one coat of inte-
rior latex primer  and one coat of semi-
gloss interior latex paint. Twelve  experi-
mental runs were conducted,  three tests
for each of the following  configurations:
empty test chamber, test chamber with
carpeting, test chamber with carpeting and
drapes, and test  chamber with carpeting
and a full-size  bed covered with a com-
forter. The carpet  pile was made of 100%
polyester fibers. The drapes were made
of 72% rayon  and 28% polyester. The
mattress consisted of 45% cotton felt, 40%
polyurethane foam, 15% polyester fibers
on a wire spring unit. The box spring con-
sisted of a wire spring unit covered with
100% polyester fabric. The mattress was
covered with a comforter consisting of 50%
cotton, 50% polyester fabric and filled with
,100,%  polyester ,fjberfi|L._T,he,,carpeting,
drapes, and bed were replaced after each
experimental run.
   Real-time temperature,  humidity, and
barometric  pressure measurements were
collected during  all experimental runs.
Temperature and relative  humidity were
controlled; barometric pressure was moni-
tored  but not controlled. Air mixing was
achieved using two commercially avail-
able box fans.  Air exchange rates within
the test chamber  were controlled for ap-
proximately  1   exchange per hour, and
monitored  using  the tracer gas decay
method.
   Each experimental run began by simul-
taneously  releasing PDCB vapor and a
tracer gas, sulfur hexafluoride  (SF6) into
the test chamber at constant  emission
rates. The flow  of SF6 into the test cham-
ber from a compressed gas cylinder was
  controlled by a fine-metering valve, and
  adjusted to maintain the air concentration
  inside the test chamber to below 500 ppb.
  PDCB  vapor was generated in  the test
  chamber by  sublimation of pure PDCB
  crystals placed in a flat metal pan that
  was positioned at the air inlet.
    A continuous sample of  test chamber
  air was collected  at the outlet. After both
  compounds were  maintained at a steady-
  state equilibrium for a 6-hr period, injec-
  tion was stopped. Air monitoring for PDCB
  and SF6, using a gas chromatograph (GC),
  began  prior to  the introduction  of  both
  compounds into the test chamber and was
  continued throughout the  air concentra-
  tion buildup period, a  6-hr steady-state
  equilibrium period, and the period of de-
  cay (after removal of the sources)rThe air
  monitoring ended  when concentrations of
  both compounds fell below their respec-
  tive detection limits.
    The air concentrations of SFe and PDCB
  measured at the  time the sources were
  removed were the  initial concentrations
  (C0) used in the decay model (Eq. 2) to
  approximate a decay curve which is rep-
  resentative of decay  caused solely by
  dilution. The  concentrations of SF6 and
  PDCB measured from the time the sources
  were removed were plotted against time,
  and curves were  fitted to the  data. The
  empirical decay curves for each chemical
  were compared to the  theoretical decay
  curves  predicted by Eq. 2. Given that Eq.
  2 provided an  adequate fit to the SF6
  data, sink effects were indicated by an
  observed rate of decay for PDCB which
  deviated  from that in  Eq. 2. This is ex-
  plained by the fact that other sources than
  the  PDCB concentration at time t = 0 (C0)
  are  contributing to PDCB concentrations
  (Ct) in  the test chamber air during the
  decay  period. This would  result in the
.. measured air concentrations  exceeding
  those predicted  by Eq. 2. The most likely
  source is PDCB absorbed  into surfaces
  during the experiment, which subsequently
  desorbs and  reenters the air. No other
  source  was identified.
    The data used in modeling the decay of
  SPe and  PDCB  concentrations over time
  were restricted to the data starting with
  the  first data point obtained just prior to
  the  withdrawal of the  gas  sources from
  the  test chamber  (see  data in Appendix
  G).  Further, the  data  were truncated at
  the  end of each run.  Data following the
  first occurrence when the SF6 concentra-
  tion fell below 5 ppb were discarded from
  the  statistical analysis. A 1 -ppm cutoff for
  runs 1 through 6 and a  0.3-ppm cutoff for
  runs 7  through  12 were used  for PDCB
  concentrations. These figures correspond
  to the  limits  of quantitation for the two
  compounds.

-------
   Data from each run were analyzed sepa-
 rately. A  series of decay  models were
 fitted to the SF6 and  PDCB concentration
 data over  time. These  models included
 first- and second-order exponential decay
 models for both compounds. A segmented
 model, consisting of a second-order expo-
 nential decay curve  followed by a first-
 order linear equation, was also fitted to
 the PDCB concentration. Each model was
 forced through  C0, the concentration cal-
 culated at time t=0 when the source was
 removed.  The equations considered take
 the general forms:

   First-order exponential decay:
     Cone = C0  exp(-At)             (3)
   Second-order exponential decay:
' """ Cone = C0  exp(-Bf + Ct2) ' "  '   (4)
   Segmented second-order exponential
      decay and first-order linear:
     Cone = C0  exp (~Dt + Et2)
      If t < T, the time of model transition,
     Cone = F + Gt
      otherwise     ,             (5)

 where A,  B, C, D, E,  F, G, and T  are
 model parameters (coefficients) to be esti-
 mated from the data. The coefficient A in
 Eq.  3, obtained for SF6, provides an esti-
 mate of the air exchange rate within  the
 test  chamber. To model Eqs. 3 or 4 above,
 the  concentrations were first log-trans-
 formed and the regression parameters es-
 timated on the  log-scale. To use a stan-
 dard linear regression analysis procedure,
 the response variable  modeled versus time
 was [log(conc) - log(C0)],  selecting  the
 no-intercept model option. The regression
 coefficients B  and C were directly  ob-
 tained from that regression analysis.
   For each individual run, the sink effect
 was estimated  as a function of time,  the
 initial PDGB concentration at time t =  0
 (C0), the  air exchange  rate  (A) of SF6
 during the same run, and the parameter
 estimates (B and C or D, E, F, G,  and T)
 of the PDCB decay function. If Eq.4 above
 best described the PDCB  concentration
 decay curve, then the sink effect, at time
 t,  expressed in ppm, was  estimated by
 the following difference:

   Estimated sink effect =
      C0 exp(-Bt + Ct2) - C0 exp(-At).

   If  Eq. 5 best described the PDCB con-
 centration decay curve, then the sink ef-
 fect, at time t, was estimated as follows:

   Estimated sink effect =
      C0exp(-Dt+Et2)-C0exp(-At)
        if t < T,
  Estimated sink effect =
     F + Gt - C0 exp(-At)
        otherwise.

Results and Discussion
  Sink effects were observed  in all  ex-
perimental runs; that is, PDCB air concen-
trations  measured  after  the  source was
removed exceeded that  predicted by  the
simple dilution  model.  The  most likely
source of excess  PDCB was from  the
chemical's adsorption to the walls, drapes,
carpeting, bed,  and comforter during  the
experiments and subsequent desorption
and reentry into the air. No other possible
source was identified.
  The maximum differences between
PDCB air concentrations predicted from
the models fitted to the calculated con-
centrations and those  of the air concen-
trations  predicted from Eq. 2 are as  fol-
lows:

  Empty test chamber        0.9 ppm
  Test chamber with carpet    1.6 ppm
  Test chamber with carpet
    and drapes              7.9 ppm
  Test chamber with carpet
    and bed                 2.6 ppm

  The PDCB decay models from each
run were also used to predict the number
of air exchanges needed  to reduce PDCB
air concentrations from  a high to  a  low
level. As an example, the models fitted to
the PDCB data in the presence of sinks
and the model described by  Eq. 2 were
used to calculate the  number  of air  ex-
changes per hour (ACH) required to  re-
duce the air concentration of PDCB from
14ppm to 2ppm (Tablel). Under the  as-
sumption of no sink effect (Eq. 2), it would
require  1.95 ACH to  reduce the PDCB
concentration from  14 to 2ppm, regard-
less of chamber configuration. (The num-
ber 1.95 is obtained by solving  the equa-
tion,  2  = 14e'At, for At.)  The average
number  of air exchanges necessary to
reduce the PDCB air concentration within
the empty test chamber from 14 to 2 ppm,
as calculated from  the decay models, is
2.50 ACH. It would require an average of
3.21 ACH for the test  chamber with car-
peting, 6.64 ACH for the test chamber
with carpeting and drapes, and  3.75 ACH
for the test chamber with carpeting and a
bed.

Summary and Conclusions
  The project report presents the data
from experimental runs using a chemical,
p-dichlorobenzene, in a room-size test
chamber. Triplicate experimental runs were
conducted in the test chamber  using four
configurations:  empty test chamber; test
chamber outfitted with wall-to-wall carpet;
test chamber with carpet and drapes; and
test chamber with  carpet and a full-size
bed with a comforter.
  Sulfur hexafluoride was used  to deter-
mine the air exchange rates of  each run
within the chamber and to demonstrate
the actual decay of the chemical in the
absence of sink effects under the environ-
mental conditions set for each  run. The
empirical decay curve for SFe,  obtained
from fitting SF6 concentrations over time
after the source of the gas was withdrawn
from the chamber, coincided with  the theo-
retical decay curve predicted by the simple
dilution model (Eq.  2).
  Sink effects were observed when the
estimated rate of decay of PDCB deviated
from the rate predicted'by Eq. 2,  using the
air exchange rate determined by  SF6. This
effect can be explained by the presence
of sources other than C0, the concentra-
tion of PDCB in the air at the beginning  of
the decay period, which are contributing
to PDCB concentration (C() in the cham-
ber during the decay period. This results
in measured PDCB air concentrations that
exceed  the air concentrations predicted
by Eq. 2 (without sink effects).  The sink
effect was found in all experimental runs.
The  most likely  source  was PDCB
adsorbed to chamber surfaces during the
experiments, with PDCB subsequently de-
sorbing and reentering the air. The  mag-
nitude of the sink effect was influenced by
furnishings added to the chamber.
  In the presence of sinks, a higher num-
ber of air exchanges was required to re-
duce PDCB air concentrations within the
test chamber to background levels after
the source of the contaminant was with-
drawn. As an example, the models  fitted
to the PDCB data in the presence of sinks
and the model  described by Eq. 2  were
used to calculate the  number of air ex-
changes required to reduce the air con-
centration of  PDCB from  14 ppm  to  2
ppm. The first-order decay model (Eq.  2)
would significantly underestimate  the num-
ber of air exchanges necessary to reduce
air concentrations to background levels.
As shown in  this study, the sorption  of
PDCB onto chamber surfaces and its sub-
sequent desorption  was significant rela-
tive to the PDCB levels in the room air.
Therefore, sink  effect terms should be in-
corporated into  predictive exposure  mod-
els. Additional modeling of the data base
generated during this study should be un-
dertaken. Fitting other models  to these
data, based on some  type of equilibrium
adsorption phenomenon or other physi-
cally-based concept, should be attempted.
                                                                                     •U.S. Government Printing Office: 1992— 646-080/60140

-------
Table 1.   Number of Air Exchanges Needed to Reduce PDCB Concentrations from 14 to 2 ppm
 Run
 no.
Time to
14 ppm
  (V
Time to
2 ppm
  (h)
  Time
difference
   (h)
Number of air
 exchanges
w/o sink effect
Number of air
 exchanges
 w/sink effect
Empty tost chamber
  1         0.65          3.13
  2         0.28          2.13
  3         0.36          2.41
Test chamber with carpet
  4         0.47          3.25
  5         0.31          2.86
  6         0.62          3.73
Test chamber with carpet and drapes
  7         0.63          5.99
  8         0.05          4.91
  9         1.00          6.40
Test chamber with carpet and bed
  10        0.62          4.07
  12        0.26          2.67
                             2.48
                             1.85
                             2.06
                             2.78
                             2.55
                             3.11
                             5.36
                             4.87
                             5.40
                             3.45
                             2.41
                             1.95
                             1.95
                             1.95
                                                         Avg. =
                                                         Std  =
                                                         CV(%)
                             1.95
                             1.95
                             1.95
                                Avg. =
                                Std  =
                                CV(%)
                             1.95
                             1.95
                             1.95
                                                         Avg. =
                                                         Std  =
                                                         CV(%)
                             1.95
                             1.95
                                                         Avg. =
                                                         Std  =
                                                         CV(%)
                               2.51
                               2.53
                               2.45
                               2.50
                               0.04
                               1.53%
                                3.27
                                3.22
                                3.14
                                3.21
                                0.07
                               2,14%
                                6.84
                                6.03
                                7.05
                                6.64
                                0.54
                               8.13%
                                3.99
                                3.51
                                3.75
                                0.34
                               9.05%
  The information  in this document has
been funded wholly or in part by the United
States Environmental  Protection  Agency
under  Contract 68-DO-0137 to Midwest
Research Institute. It has been subjected
to the Agency's  peer and administrative
review, and it has been approved for pub-
lication as an EPA document.
  Mention of trade names or commercial
products does not constitute endorsement
or recommendation for use.
  The EPA authors are James Quakenboss, Environmental Monitoring and Systems
    Laboratory, Las Vegas, NV 89193-3478, and Janet Remmers, Office of Pollution
    Prevention and Toxics, Washington, DC 20460. James McHugh andKarin Bauer are
    With the Midwest Research Institute, Kansas City, MO 64110.
  Joseph V. Behar is the EPA Project Officer (see below).
  The complete report, entitled 'Tests of Indoor Air Quality Sinks," (Order No. PB92-
    218346/AS; Cost: $43.00; 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:
          Environmental Monitoring and Systems Laboratory
          U.S. Environmental Protection Agency
          Las Vegas, NV 89193-3478
United States
Environmental Protection Agency
Center for Environmental Research Information
Cincinnati, OH 45268

Official Business
Penalty for Private Use
$300
                                                                                     BULK RATE
                                                                                POSTAGES FEES PAID
                                                                                         EPA
                                                                                   PERMIT No. G-35
EPA/600/SR-92/094

-------