United States
                  Environmental Protection
                  Agency
Air and Energy Engineering
Research Laboratory
Research Triangle Park, NC 27711
                  Research and Development
EPA/600/SR-93/239   March 1994
x EPA      Project Summary
                  Characterization  of  Air
                  Emissions  from  the  Simulated
                  Open  Combustion of  Fiberglass
                  Materials
                  Christopher C. Lutes and Jeffrey V. Ryan
                    The exposure of persons  to fiber-
                  glass combustion emissions from
                  structural fires, fires at waste landfills,
                  and fires at demolition sites has be-
                  come an issue of increasing  concern.
                  This  study  identifies and quantifies a
                  broad range of pollutants that are dis-
                  charged  during the small-scale, simu-
                  lated, open  combustion of fiberglass
                  and reports these emissions relative to
                  the mass of fiberglass material com-
                  busted. Two types of fiberglass materi-
                  als (representing the boating and build-
                  ing materials industries) were com-
                  busted in a  controlled outbuilding de-
                  signed for the simulation of open burn-
                  ing. Volatile, semivolatile, and particu-
                  late-bound organics were collected and
                  analyzed by gas chromatography/mass
                  spectrometry. The emphasis of these
                  analyses was on the quantification of
                  hazardous air pollutants listed in  Title
                  III of the Clean Air Act Amendments of
                  1990, although further efforts  were
                  made to identify  and quantify other
                  major organic components. Additional
                  sampling and analysis were  done for
                  hydrogen chloride, particulate-phase
                  metals,  and respirable fibers. Fixed
                  combustion gases (carbon dioxide, car-
                  bon monoxide, nitric oxide, oxygen, and
                  total hydrocarbons) were monitored
                  continuously throughout the test pe-
                  riod. Analytical results show substan-
                  tial emissions of a  large number of
                  pollutants including arsenic, benzene,
                  benzo(a)pyrene,  carbon monoxide,
                  dibenzofuran, lead, naphthalene,  par-
                  ticulate,  phenanthrene, phenol, styrene,
                  and toluene.
  This Project Summary was developed
by EPA's Air and Energy Engineering
Research  Laboratory, Research Tri-
angle Park, NC, to announce key find-
ings of the research project that is fully
documented in a separate report of the
same title (see Project Report ordering
information at back).

Introduction
  Concerns over exposure to air emis-
sions from the open burning of fiberglass
in structural fires, and  at waste disposal
and demolition sites have been expressed
to EPA's Control Technology Center by
government entities including the state of
Alaska. Though little previous  research
has been done specifically on combustion
emissions from fiberglass,  literature does
exist relating to the composition of fiber-
glass, the combustion  products of some
components of commercial fiberglass ma-
terials, and the suspected health effects
of fiberglass fibers. Fiberglass  is  princi-
pally composed of SiO2 (approximately
50% by weight); additional major compo-
nents are AI2O3, Ba2O3, CaO, and MgO
(typically 3 to 20% each); and trace com-
ponents include F, Fe2O3, K2O, Na2O, SO3,
and  TiO2 (less then 1% by weight each).
Additionally, fiberglass materials may con-
tain  organics  as  sizings or binders. Or-
ganics-containing fiberglass materials can
be classified as either epoxy- or polyes-
ter-based.  The known combustion prod-
 ucts of polyester-based  materials have
 been reviewed and are known to include
 acetaldehyde, benzene, biphenyl, carbon
 monoxide, ethyl benzene, pentadiene, sty-
 rene, and toluene. Much of the available
                                                                L(A) Printed on Recycled Paper

-------
information on these combustion products
has been obtained in small-scale studies
of materials that are likely to be less com-
plex than the commercial materials found
in practice. Therefore, larger  scale tests
of complex  commercial  materials under
simulated open  combustion  conditions
promise increased insight.
  In addition to  potential hazards of a
chemical nature, the air  emissions from
open fiberglass combustion may include
fibrous aerosols; the physical nature of
fibrous aerosols  may  lead to additional
health  hazards.  Epidemiological studies
have shown significant increases in non-
malignant respiratory disease in popula-
tions exposed to glass fibers. Glass fibers
are  apparently less harmful to  health than
asbestos fibers. The greatest  hazard  ap-
pears to be  related to  fibers less than
1.5  (im in diameter and longer than 8 u,m.
No measurements of fiberglass fiber emis-
sions  from combustion  processes  have
been found in literature reviewed to date.
An  assessment of  the concentration and
size distribution of fibrous aerosols  pro-
duced  from fiberglass open combustion
processes would  be valuable.
  In response, with the guidance of EPA's
Air  and  Energy Engineering  Research
Laboratory  (AEERL), a study was under-
taken to measure emissions from the com-
bustion  of  fiberglass samples from two
industries that use  fiberglass extensively.
This study  included replicate  tests of fi-
berglass materials from the boating indus-
try (polyester-based, some with and some
without a gel coating—a  colored sealant),
and the building  industry  (vinylester-
based). The study was  designed to col-
lect, identify, and quantify a wide range of
air  emissions and  to report these emis-
sions per mass of fiberglass material com-
busted. The emphasis of these analyses
was on the quantification  of air toxics com-
pounds listed in Title III  of the Clean Air
Act Amendments of 1990 (CAAAs), al-
though further efforts were made to iden-
tify  and semiquantify other major organic
components.

Methods
  Combustion testing for  this study  oc-
curred  in EPA's Open Burning Simulation
facility. This facility consists  of  a 2.7 by
3.4  m  outbuilding  equipped  with a  plat-
form scale, air  handles,  and  sampling
equipment  used  for combustion studies.
The boating industry fiberglass samples
(a total of 8-10 kg per test) were placed in
the  facility and ignited using a brief appli-
cation of a hand-held propane  torch which
was removed before sampling began. A
"hut blank" test, in which the propane torch
was briefly introduced into the facility but
no  fiberglass was  combusted, was  con-
ducted for comparison. In order to allow
adequate time for all necessary samples
to be obtained, three separate charges of
fiberglass  were combusted during each
test. Combustion of one  charge  was al-
lowed to go to apparent completion (as
signified by constant weight and near back-
ground  concentrations  of combustion
gases) before  another charge was intro-
duced. Attempts to test the building indus-
try fiberglass sample in like manner were
unsuccessful because of the high concen-
tration of flame retardant in this  sample.
Therefore, the  combustion of the  building
industry material (a total of 4 to 6 kg per
test) was supported by a continuous liquid
propane (LP) gas flame during sampling.
This study design was  intended to simu-
late the behavior of this  flame-retarded
fiberglass material in the presence of other
non-flame-retarded combustibles.  A "com-
bustion blank' test, in which the LP flame
was operated  but  no fiberglass  was
present, was conducted for comparison.
In addition,  various field and  laboratory
blank samples  were collected for each
sampling train. In order to allow adequate
time for all necessary samples to be ob-
tained, two separate charges of fiberglass
were combusted during each test. Com-
bustion of one charge was allowed to go
to apparent completion, as was done for
the  boating industry samples, before an-
other charge was introduced.
  An elemental analysis of the fiberglass
samples  was  performed  before  testing.
Fixed combustion gases  [carbon dioxide
(CO2), carbon monoxide (CO), nitric oxide
(NO), oxygen  (O2),  and  total hydrocar-
bons (THC)] were monitored continuously
throughout the test period through the sam-
pling manifold. Temperatures  at  relevant
locations in and around  the test facility
and the mass of fiberglass material were
monitored throughout the  test period.
  Volatile  organics were  collected  from
the  sample manifold on  sorbent  tubes
(VOST train) and analyzed by gas chro-
matography/mass spectrometry (GC/MS).
Since extremely high levels of volatile or-
ganic compounds were observed during
early tests of the building industry fiber-
glass,  an  additional volatile sampling
method  was implemented. Samples were
collected in Tedlar® bags and analyzed by
GC/MS. Samples for semivolatile and par-
ticulate-bound organics and metal aerosol
analysis were collected through separate
medium volume PM10 samplers in the burn
hut. The sample for metal aerosol was
collected on a 142-mm-diameter quartz
fiber filter  and analyzed  by  inductively
coupled  plasma-atomic emission  spec-
trometry. The semivolatile and particulate-
phase organic sample was collected with
a 142-mm-diameter,Tef lon®-impregnated,
glass fiber filter and XAD-2® resin sorbent.
The filter and resin were then extracted in
methylene chloride, and the  pooled  ex-
tract was analyzed by GC/MS. A real-time
photoelectric analyzer, designed to  quan-
tify total polycyclic aromatic hydrocarbons
(PAHs) on submicron particulate, was also
operated using a sample stream withdrawn
through the  sampling manifold. A sample
probe for vapor-phase  hydrogen chloride
was  also located  in the burn  hut. These
samples were analyzed by ion chroma-
tog rap hy.
  The sampling train used to sample for
fiber size and morphology analysis con-
sisted of a 37-mm mixed cellulose ester
filter cassette  followed  by a low volume
sampling pump and dry gas  meter. The
filter was operated in an inverted position,
parallel to the facility floor during sampling
to minimize the collection of  particulate
matter through gravitational settling. Analy-
sis was performed by phase contrast light
microscopy (PCM) and  transmission elec-
tron  microscopy (TEM). For these analy-
ses,  a fiber was defined as a particle with
an aspect ratio of  greater than 3:1. These
sampling and analysis methodologies were
based  on NIOSH methods for asbestos
fibers.
  After  the  completion of the chemical
and  microscopic  analyses, analyte con-
centration data were coupled with sample
volume, facility air flow, and  combustible
material  mass  loss data to  derive esti-
mated  emissions  (expressed as mass of
analyte  produced  per  mass of fiberglass
material consumed in the combustion pro-
cess).

Results and Discussion
  The elemental analysis of the fiberglass
materials before combustion (Table 1) in-
dicates that  the organic matter content of
the boating  industry fiberglass is higher
than that of building industry material. The
substantial halogen concentration  found
in the building industry material tends to
confirm  the manufacturer's statement that
the material contained  a brominated  fire
retardant. The vast majority  of the com-
bustion of each charge of boating industry
material was completed in a 20-40 minute
period, while the  majority of the building
industry material in each charge appeared
to be consumed in 30-60 minutes.  Table
2 summarizes the  estimated  emissions
derived  from real-time  measurements of
CO,  CO2, THC, and total  PAH bound to
submicron particulate. The substantial CO

-------
emissions observed,  which are probably
underestimated in this data set, are a con-
cern since CO is believed to be the most
important cause of death of fire victims.
Substantial emissions of particulate  mat-
ter  were also observed  (average values
were  117 g/kg  for the  boating  industry
material  and  607 g/kg for the building in-
dustry material). This is  a concern since
most previous studies of combustion prod-
ucts of various polymers have paid  little
attention to the composition of the particu-
late phase.
  The volatile organic data  set includes
concentration measurements for  35  tar-
geted  (the  majority of which  are  consis-
tently non-detectable) and several dozen
tentatively identified species; Table 3  pre-
sents  average data  on  several selected
compounds that were among those seen
in the highest levels.
  It appears that the relative ratios of these
components  are similar for both the boat-
ing and building industry materials but that
the absolute  emission rate is higher for
the building industry material.
  The semivolatile and particulate bound
organics data set generated from  this
project includes  concentration measure-
ments for  more  than 90  targeted com-
pounds (the  majority of  which were con-
sistently  non-detectable)  and  several
dozen tentatively identified species. Aver-
age emission values  for  a selected set of
detected, targeted semivolatile, and  par-
ticulate bound organics  are  presented in
Table 4. Average estimated emissions for
these compounds are generally lower than
for  the volatile species  discussed previ-
ously. As in  previous measurements,  the
values  obtained  in the  building industry
fiberglass tests are generally higher than
those  in the   boating  industry tests.  Pre-
liminary  calculations  have shown that the
estimated emissions calculated from  the
output of the  real-time PAH analyzer (Table
2) agree at least within a factor of 10  with
the sum of  estimated emissions  calcu-
lated  from  the Method 8270  analyses of
PAHs that would be  expected to  be  pre-
dominantly in the particulate  phase.
   Fibrous aerosols samples rarely showed
significantly more fibers than were seen in
blank samples, as shown in Table 5. How-
ever, detection limits  were quite high for
this analysis, since the maximum feasible
loading of total particulate on these filters
was reached after a very  small  volume
(<20  L) was  sampled, and it is feasible to
conduct the microscopic examination  only
on  a small representative portion of  the
filter surface  area.
  The particulate-phase  metals  samples
were analyzed for 11  elements. Of these,
Table 1.   Composition of Fiberglass Materials Tested (All Data as Percent Composition)

                                                         Building
                                                         Industry
                                                         Fiberglass

                                                           2533
                                                            2.48
                                                           10.06
                                                           <05
                                                           <0.5
                                                            1.9
                                                            2.2
                                                            0.12
                                                           <0.01
                                                            0.086
                                                            0.52
                                                           35.29
                                                           60.23
                                                            3.96
Boating Industry
Fiberglass -
Without Gel Coat
Carbon
Hydrogen
Oxygen
Nitrogen
Sulfur
Total Halogen (as Cl)
Aluminum
Magnesium
Cadmium
Chromium
Moisture
Volatile Matter
Ash
Fixed Carbon
52.97
4.79
7.1
0.017
<0.03
<0.5
0.28
<0.05
<0.004
0.0091
2.06
63.11
34.83
<0.1
Boating Industry
Fiberglass -
With Gel Coat
55.06
5.27
<0.5
0.015
<0.03
<0.5
0.081
<0.05
<0.004
0.041
1.19
56.24
39.57
3
Table 2.
  Combustion Gas and PAH Particulate Analyzer Concentrations and Estimated
  Emissions
                 Test
09/08/92
09/17/92
10/13/92
10/30/92
11/04/92

~Key:  ~
    First Boating Industry Test
    Second Boating Industry Test
    First Building Industry Test
    Second Building Industry Test
    Third Building Industry Test

CO as C
Estimated
Emissions
(g/kg)
•t 48.2*
Test 55.2*
;f 205.9"
Test 141.4*

-------
Table 4. Estimated Emissions of Selected Semivolatiles
Boating Industry Building Industry
Compound Average (mg/kg) Average (mg/kg)
Anthracene 353 202
Benzo(a)pyrene 86 72
Biphenyl 689 1,936
Chrysene 323 458
2-Methyl Phenol 125 400
Phenanthrene 902 2,156
Phenol 328 6,830
Table 5. Fibrous Aerosol Measurements
PCM Length >5 ATEM Length >0.5 ATEM Length
usn Estimated <5.0 /iim Estimated >5 /MTI Estimated
Sample Test Emissions Emissions (million Emissions (million
No. Date Test (million S/kg) S/kg) S/kg)
7 09/17/92 Second Boating Industry 1710 490*# 420*#
8 09/17/92 Second Boating Industry 4846* 1902*# 1902*#
11 09/17/92 Second Boating Industry 504* NFD*# NFD*#
16 10/13/92 First Building Industry 1904* 395* 395*
17 10/13/92 First Building Industry 432* 89*# 89*#
19 10/30/92 Second Building Industry 231*# 120*# 120*#
Key: NFD = No libers detected, detection limit cannot be stated accurately due to
loading problems.
PCM = Phase contrast microscopy.
ATEM = Analytical transmission electron microscopy.
= Number of observed fibers not greater then 3 times larger than the appli-
cable field and hut blank values.
# = Air concentration of observed fibers not more then 3 times larger than the
applicable hut or combustion blank value
S = Fibrous structures.
phases of the combustion process. In ad-
dition, since the rate of emissions from a
small mass of combusted fiberglass was
high enough to threaten overloading of
the sampling media, it was necessary to
sharply limit the amount of fiberglass com-
busted in each test phase. This may in-
troduce a significant source of error into
the data set because the resolution of the
balance used to measure the weight of
fiberglass lost to combustion was 0.09 kg
(0.2 Ib): (the average weight losses for
the samples were as follows: organic
semivolatile/particulate train 1.4 kg (3.0
Ib), metals train 1.7 kg (3.8 Ib), VOST
train 0.4 kg (0.9 Ib), Tedlar® bag train
0.91 kg (2.0 Ib), GEM train 3.8 kg (8.3 Ib),
fiber train 0.45 kg (1 Ib), and hydrochloric
acid train 3.3 kg (7.2 Ib)).
Conclusion
Despite the aforementioned experimen-
tal difficulties, this project did succeed in
producing estimated emissions data for a
broad range of atmospheric pollutants from
simulated open fiberglass combustion.
Substantial emissions of a large number
of pollutants were observed, including ar-
senic, benzene, benzo(a)pyrene, carbon
monoxide, dibenzofuran, lead, naphtha-
lene, particulate, phenanthrene, phenol,
styrene, and toluene.
Table 6.  Metals Estimated Emissions (mg/kg)

Test                        Silver     Arsenic
                                  Cadmium
First Boating Industry
Second Boating Industry
Boating Industry Average
First Building Industry
Second Building Industry
Third Building Industry
Building Industry Average
4.41
4.60
4.50
13.38*
14.45*
6.76*#
11.53
<0.52*#
<0.6*#
0.56
6.13*
48.15
13.05
22.44
0.06*#
0.29
0.18
<0.28*#
<0.80*
<0.24*#
0.44
                                                Chromium    Lead
1.57*#
0.48*#
1.03
8.92
14.45*
5.80*#
9.72
21.00
38.72
29.86
<2.79*#
<8.03*
<2.42*#
4.41
Key:
Mass of this sample not greater than 3 times the largest of the following: mass in
field blank, mass in hut blank, or mass in combustion blank, as applicable.
Sample hut air concentration not greater than 3 times the hut blank air concentra-
tion or the combustion blank air concentration, as applicable.
                                                                                      . GOVERNMENT PRINTING OFFICE: IW4 - 550-067/80203

-------

-------

-------

-------
   C. C. Lutes and J. V. Ryan are with Acurex Environmental Corp., Research Triangle
    Park, NC 27709.
   Paul M. Lemieux is the EPA Project Officer (see below).
   The complete report, entitled "Characterization of Air Emissions from the Simulated
    Open Combustion of Fiberglass Materials," (Order No. PB94-136231;  Cost:
    $19.50; 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
          Air and Energy Engineering 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/600/SR-93/239

-------