-------

-------
4814
TIRES, OPEN BURNING
combustion of oil, coal, or wood in utility boilers; it is most
similar to values for the open burning of wood or plastic. These
results demonstrate that the open burning of scrap  rubber
tires produces an extremely high mutagcnic emission factor,
posing a potential environmental and health hazard.
BACKGROUND

Open Burning of Tires

Approximately 240 million scrap rubber tires (2.2 million tons)
are discarded  annually in the U.S. (1,2). Viable methods for
reclamation exist. Some  of the attractive options for use of
scrap tires include burning, either alone or with another fuel,
such as coal, in a variety of energy-intensive processes, such
as cement kilns and utility boilers (3-5). Another potentially
attractive option is the use of ground tire material as a sup-
plement to asphalt paving materials. Congress has passed a
law, the Intermodal Surface Transportation  Efficiency Act
of 1991 (6), which mandates  that up to 20% of all federally
funded roads  in the U.S. include as much as  20 Ib (9  kg) of
rubber derived from scrap tires per ton (907 kg) of asphalt
by 1997. In spite of these efforts, less than 25% of the total
amount of discarded tires are reused or reprocessed, and the
remaining 175 million scrap tires are discarded in landfills,
above-ground  stockpiles,  or illegal dumps. In addition, these
reclamation efforts do little to affect the estimated 2 billion
tires (18 million tons) already present in stockpiles.
   Many landfills no longer accept scrap tires because  of the
disposal and  health-related problems posed  by used  tires.
After  burial,  tires often float to  the surface  and become
partially filled with water, which serves as an ideal breeding
habitat for many insects, especially mosquitoes (7). In fact,
the introduction and spread of several mosquito species  in the
U.S. have been attributed directly to the formation of breeding
habitats by water-filled tires  in tire stockpiles (8). Although
cutting tires in half or shredding them into pieces can reduce
the tendency of scrap tires to accumulate water, such processes
are costly, and many landfills lack the necessary equipment.
   Another problem associated with scrap rubber tires  is the
frequent occurrence of tire fires at tire stockpiles. These fires,
which are often started by arsonists, generate large amounts of
heat and smoke and are difficult to extinguish. This is partly
due to the fact that tires, in general, have more heat energy
by weight than does coal (37,600 kJ/kg vs 27,200 kJ/kg) (5).
Some tire fires have burned continuously for months, such as
the 9-month Rhinehart tire fire in Winchester, Virgina. Such
fires pollute not only  the atmosphere  but also the land and
groundwater due to the liquefaction of the rubber during the
combustion process.
   Considerable research has been performed  on the  muta-
genic and carcinogenic properties of chemicals associated with
the rubber industry (9,10). These studies have shown that a
wide variety of mutagens and carcinogens are present in the
rubber industry, and  that cancer and other types of health
effects have been associated with rubber workers (9).  Many
chemicals, some of which are mutagens and carcinogens, have
been identified in the emissions from both the controlled and
uncontrolled burning of rubber tires (4,11);  however, we are
unaware of any studies on the mutagenicity of the emissions
produced by the open burning of tires.
                                                    The organic products of incomplete combustion (PICs) are
                                                 present in emissions from all combustion processes and, in gen-
                                                 eral, have been found to be carcinogenic in humans and rodents
                                                 and to be mutagenic in bacteria and mammalian cells (12). Mu-
                                                 tagenicity bioassays,  especially the Salmonella mutagenicity
                                                 assay, have been shown to be useful for evaluating the health
                                                 effects of airborne mutagens and potential carcinogens present
                                                 in the PICs from a variety of combustion emissions (13). Such
                                                 bioassays have been used to characterize PIC-impacted urban
                                                 air as well as emissions from the combustion of municipal and
                                                 hazardous waste, polyethylene plastic, woodsmoke, and auto-
                                                 motive and diesel exhaust (14-17).

                                                 Sources of Data

                                                    EPA 1989 Laboratory Tests. In 1989, the EPA's Air and En-
                                                 ergy Engineering Research Laboratory (AEERL) performed
                                                 experiments under the auspices of the EPA's Control Technol-
                                                 ogy Center (CTC) to evaluate and characterize the emissions
                                                 from a simulated tire fire (11,18). This study involved burn-
                                                 ing small (11 kg (25 Ib)) amounts of tire material in an open
                                                 burning  test facility. Emissions of volatile  and semivolatile
                                                 organics, participate matter, and  criteria  pollutants  were
                                                 measured and quantified. Results were reported in both con-
                                                 centration units and emission factor units  based  on mass
                                                 emitted per mass of tire burned.
                                                    One of the findings of this study was that there are thou-
                                                 sands of compounds present in the emissions from a tire fire.
                                                 Although many individual compounds were identified, and a
                                                 significant fraction of the organic mass was identifiable, only
                                                 a small fraction of the total number of chromatographic peaks
                                                 were identifiable. The complexity of this mixture made it dif-
                                                 ficult to assess the  threat to human health that is posed by a
                                                 tire fire; therefore, the samples from this study were archived
                                                 for later use, in the hopes that further research could be per-
                                                 formed on the samples at a future date.

                                                    EPA Bioassay Tests. In 1992,  the EPA's Health Effects Re-
                                                 search Laboratory  (HERD, in cooperation with AEERL and
                                                 again funded by the CTC, evaluated the previously archived
                                                 samples  for biological activity,  using the Salmonella  TA98
                                                 bioassay (19,20). In addition, high performance liquid chro-
                                                 matography (HPLC) techniques were used to fractionate the
                                                 samples, and the resulting fractions were subjected to similar
                                                 bioassays. This bioassay-directed fractionation process iden-
                                                 tified the biologically active fractions of the original samples,
                                                 and gave insight into the compounds that were contributing
                                                 to the biological activity. In addition, the mutagenic emission
                                                 factors were calculated and compared to other conventional
                                                 combustion sources.

                                                    Ontario Field Samples. In the time period during which the
                                                 HERL bioassay-directed fractionation experiments were ongo-
                                                 ing, a large tire fire began. The field tire fire was initiated by
                                                 arsonists on the evening of February 12,1990, in a small sec-
                                                 tion of the Tyre King tire recycling depot located in a rural sec-
                                                 tion of southwestern Ontario, Canada. The fire spread quickly,
                                                 involving approximately 14 million scrap tires that were stored
                                                 on the site. This fire is considered to be the largest tire fire
                                                 thus far in North America (21). Dense smoke forced the evacu-
                                                 ation of all residents within a large area. The fire was com-
                                                 pletely extinguished  after 17 days of intensive firefighting,

-------
which included aerial bombardment with fire retardants. Ap-
proximately 30% (6.5 million L) of the water used to fight the
fire was retained on the site in lagoons, where it was treated by
an emergency water treatment plant. The treated water, which
met provincial standards, was  released into a local creek. An
additional 12.8 million L of water were trucked to two local
sewage treatment plants for bioremediation and disposal. The
fire produced large quantities  of oil, and approximately 95%
(1 million L) of the oil was disposed of by burning as a haz-
ardous waste in a licensed facility. The remaining oil (approxi-
mately 50,000 L) was estimated to have seeped into the ground
at the site of the fire. Personnel from McMaster University ac-
quired samples from  the plume of this tire fire and performed
chemical and biological analyses that were similar to those per-
formed by EPA's HERL. Slight differences in the composition
of the  organics between the laboratory and field samples are
possibly due  to atmospheric transformation reactions due to
ultraviolet light, because the laboratory tests were performed
indoors.

   Other Data.  There were two additional sources of data that
were considered for  use in this study, one in which ambient
monitoring data taken downwind from a number of tire fires
were analyzed (22), and one in which NIOSH performed some
limited ambient sampling at the Winchester, Va tire fire (23).
Unfortunately, the only data available in the first study were
for volatile organic compounds (VOCs) which do not condense
on the fine particulate matter typically produced from the tire
fires. The  fine particulate, with condensed (PAHs) or other
high molecular weight compounds, is of greater concern from
tire fires, due to the fine particles' ability to be inhaled deeply
into the lungs. In addition, the data taken for the first study
(22) were of variable quality, using differing (and sometimes
uncharacterized) sampling methods, making it difficult to an-
alyze and interpret.

Bioassays
A bioassay is a study in which the effects of an agent are
evaluated in an organism or a component of an organism, such
as an organ, tissue, or enzyme. Various complex environmental
mixtures, such as coal tar and  cigarette smoke, have be TI
studied for their biological effects in various bioassays lor
nearly a century. Most early bioassay studies with combustion
emissions involved the application of organic extracts of the
emissions onto the skin of rodents (rabbits, rats, or mice) with
the end point being skin tumors. These skin carcinogenesis
studies were critical  in the recognition that most combustion
emissions were carcinogenic, ie, they caused cancer, in these
laboratory-based systems.
   During  the past several decades, studies have indicated
that cancer cells contain mutations, ie, changes in the genetic
material (DNA),  in  certain genes  and that many cancer-
causing agents (carcinogens) are also mutagens (agents  that
cause mutation in DNA). Typical rodent-based cancer studies
cost $1 million and require several years to perform. However,
genetic toxicity assays (bioassays that detect the ability of an
agent to induce mutations in DNA) can be performed for just a
few thousand dollars in less than a month. Twenty-five years
of systematic study have  indicated that the use of selected
genetic toxicity assays can be highly useful and efficient in the
evaluation of combustion emissions for mutagenic and, poten-
tially, carcinogenic activity. The assay used most commonly
                             TIRES, OPEN BURNING      4815

 for this purpose is the bacterial  mutagenicity assay called
 the Salmonella  (Ames) mutagenicity assay, which was first
 introduced by Dr. Bruce Ames in 1970.
   An important application of the Salmonella assay is to
 determine the mutagenic chemical classes within combustion
 emissions by means of bioassay-directed chemical analysis.
 With this approach, the whole mixture is first tested for muta-
 genicity; then chemical fractions and subfractions are tested.
 Analytical studies  on the mutagenic fractions/subfractions
 can then be performed to identify the  chemical  classes or
 single chemicals that may account for much of the observed
 mutagenic activity.  Because of the chemical  complexity of
 most complex mixtures, including combustion emissions, it is
 difficult if not impossible to know a priori which chemicals
 to look for in  a complex mixture. Consequently, the  main
 purpose of bioassay-directed chemical analysis is to identify
 the chemical class as well as some representative chemicals
 that are responsible for most of the biological activity of a
 complex mixture.
    Unlike pure  compounds, environmental complex mixtures
 are not generally obtainable from the catalogs of chemical com-
 panies. Instead, they must be obtained from the environment
 by first selecting and then sampling the appropriate environ-
 ment. The sampling, storage, and extraction of samples are
 critical elements in generating data relevant to human expo-
 sure. The need  for engineers and chemists in planning sam-
 pling strategies can be illustrated by describing some of the
 considerations that need to be taken into account when sam-
 pling air for mutagenic activity.
    For  example, it is preferable to cool and dilute the air par-
 ticles from combustion sources, such as incinerator emissions,
 because in nature, a similar process occurs when combustion
 products are emitted from a chimney or exhaust pipe. The cool-
 ing and dilution causes the organic materials that are in the
 vapor state to condense onto the particles. It is particles of this
 type (typically less than 2.5 /tin in size) that are respired and
 that contain mutagenic compounds. Thus, particles collected
 without first being cooled and diluted may not contain many
 extractable organic compounds. Instead, the organics will have
 been lost to the  atmosphere. In addition to these concerns, one
 must also consider the rate at which air is sampled, the area
 of the filter, the time and temperature at which the sample re-
 mains on the filter, the size and direction of the opening on the
 sampler, and the engineering and chemistry parameters.
    Although a wide range of analytical chemical techniques are
 used to fractionate and identify chemical components in com-
 plex mixtures, there are a few key elements to consider when
 selecting an appropriate analytical technique to couple to a
 bioassay. First, the ability of the technique to recover both the
• mass and the biological activity of the fractions relative to the
 whole,  unfractionated  mixture is vitally  important. This can
 be examined by mixing the individual fractions to produce a
 reconstituted whole mixture. The mass and biological  activ-
 ity of this reconstituted mixture can then be compared to the
 original unfractionated mixture. Second, the mass and biolog-
 ical activity of the fractions can be summed to see if the sum
 of the biological activity of all fractions added together is equal
 to that of the unfractionated mixture. Although a variety of
 chemical interactions no doubt occur in complex mixtures, in-
 cluding additivity, antagonism, and synergism, many complex
 mixtures show additivity: ie, when the activities of the frac-
 tions are added together, the sum approximates the activity

-------
4816
TIRES, OPEN BURNING
of the whole, unfractionated mixture. However, this apparent
additivity may be masking a bewildering variety of chemical
interactions.
   One of the important contributions that bioassay-directed
chemical analysis has made to the study of complex mixtures is
the recognition that components that account for a small pro-
portion  of the mass of the mixture may, in  fact, account for
a sizable proportion of the mutagenicity of the mixture. Ex-
amples  include nitroarenes in diesel exhaust, PAHs in urban
air, and aromatic amines in cigarette smoke. A finai step in
confirming the contribution of certain chemicals to the mu-
tagenicity of a mixture is to determine the mass of various
individual chemicals in the mixture and mutagenic fractions.
Once the mass and mutagenic potency of certain individual
compounds are known, the specific mutagenic activity of each
can be calculated for the mixture. This provides an indication
of the relative contribution of each chemical or chemical class
to the mutagenic activity of the mixture or fraction.


EXPERIMENTAL

EPA 1989 Laboratory Tests

The project consisted of a parametric study to  collect organic
and inorganic emissions from the simulated open combustion
of scrap tires. Small quantities (10-20 Ib  (4.5-9.0 kg)) of
scrap-tire material were burned under two different controlled
conditions determined by the size of the material. Other pos-
sible variables that are possibly important (such as the type
of tire) were not included in the study due to cost constraints.
One size was about one-quarter to one-sixth of an entire tire
and will be referred to as the CHUNK condition. The other
size consisted of 2 by 2-in. (5 by 5-cm) pieces of tire and will
be referred to as the SHRED condition. The conditions were
evaluated in duplicate on successive days. An existing burn
hut used for similar projects was modified to accommodate
this task. A separate outbuilding housed the required organic
and particulate sampling equipment. A mobile laboratory was
used to  monitor fixed combustion gases, including oxygen (02),
carbon  dioxide (CO2), carbon monoxide (CO), total hydrocar-
bons (THCs), and sulfur dioxide (SO2). Organics were collected
using the volatile organic sampling train (VOST) (24) and a
semivolatile collection system using XAD-2 and particulate
filters. Particulate was also collected to assess airborne metals
and to  measure the amount of particulate that was sized
<10 ftm (PMio).  The organic constituents were analyzed both
qualitatively and quantitatively by gas chromatography/mass
spectroscopy (GC/MS), gas chromatography/flame  ionization
detector (GC/FID),  high  pressure  liquid  chromatography
(HPLC), and gravimetric (GRAY) methodologies.

   Open Burning Test Facility. Burn Hut. The burn hut is an
8 X 8 X 8-ft  (2.4 X 2.4 X 2.4-m)  outbuilding  modified  for
small-scale combustion  experiments (Fig. 1). The building
had been fitted with a cooled, dilution air handling system ca-
pable of delivering nominally 34.0 ms/min (1,200 ftVmin). A
16 X 16 X 16-in. (0.4 X 0.4 X 0.4-m) stainless-steel bi'rn pit
insulated with fire brick was mounted on a weigh scale to con-
tinuously monitor weight differential. A PMio ambient sampler
was located in the hut to collect particulate matter 10 /xm in
diameter or less. A deflector shield was located 1.2 m (4 ft)
over the pit to deflect flames, protect the ceiling, and enhance
                                                        Sample duct
                                                      Air inlet  I—j- Weighing platform
Air inlet
                                                              Figure 1. Diagram of burn hut.

                                                 ambient mixing. The gaseous sample duct opening was lo-
                                                 cated directly over the deflector shield. This duct transported
                                                 a representative portion of gaseous sample to the sample shed
                                                 immediately adjacent to the burn hut (Fig. 2). The duct was
                                                 insulated outside the hut to minimize heat loss and condensa-
                                                 tion of organics.
                                                   Sample Shed. The sample shed contained the majority of
                                                 the sampling equipment: the VOST system, the semivolatile
                                                 organic collection system, the airborne metals particulate col-
                                                 lection system, the continuous emission monitor (CEM), the
                                                 particulate  removal system, and the digital readout for the
                                                 weigh scale. Particulate samples were collected by a PMio
                                                 ambient sampler containing Teflon-impregnated glass-fiber
                                                 (TIGF) filters (Pallflex, No. T60A20,  20.3 X 25.4 cm). All
                                                 gaseous samples were extracted from  a sampling manifold
                                                 within the duct. Figure 3 diagrams the individual sampling
                                                 systems and illustrates how each obtained a representative
                                                 sample from the duct. Volatile and semivolatile organics, met-
                                                 als, and PMjo were collected and analyzed as described in the
                                                 previous study (11).
                                                   Mobile Laboratory. The mobile laboratory was a modified
                                                 recreational vehicle containing all necessary equipment to per-
                                                 form emissions monitoring on stationary combustion sources,
                                                 including CEMs, online GCs, VOST, and Modified Method 5
                                                 (MM5) sampling capabilities. The laboratory contains a sam-
                                                 ple preparation area, and is fitted with a microcomputer-based
                                                 data acquisition system. A heated sample line was connected
                                                 from the particulate conditioning filter to the sample manifold
                                                 in the vehicle. A portion of the heated sample was routed to the
                                                 S02 and THC analyzers. The remaining portion of the sample
                                                 stream was further conditioned (moisture removal by a refrig-
                                                 eration condenser and silica gel) before being routed to the 02,
                                                 C02, and CO analyzers.

                                                   Volatile Organics. Volatile organics were collected using an
                                                 unmodified VOST  system operated according to Method 0030
                                                 found in SW-846 (24). During this study, no stack probe was
                                                 used. An insulated section of 0.6 cm (1/4-in.) Teflon tubing
                                                 affixed to the sampling manifold was used to transport the
                                                 gaseous sample from the sample duct to the VOST system.

-------
                                                                                     TIRES, OPEN BURNING     4817
     Mobile laboratory
Sample shed
                                                             Burn hut
                      Heated
                                                   Insulated
^•*-
r


i

.-— • — ~ ^-—
^- 	 -^
OEMs 0,
C02
CO
THC
S02
I =====

1 EPA 1
1
»•«.
•^
r

^s.
D
sample
line
A

j
1 1
Sampling
Particu
Volatile, a
organ
Airborne
control c<
late sampl
nd semivo
ic samplir
metals san
o
inter:
ing
latile
g
ipling
sam
du
/I
H

pie
ct
X \


V
                                                                             Figure 2. Location? of required test
                                                                             equipment.
                                           Duct cross section
                                                                                      From burn hut
                                                             VOST system
                                                             condensers,
                                                            meters, pumps
                                         142-mm
                                         Teflon-
                                         coated
                                         filter
                                    Figure 3. Diagram of sampling systems used.
The sample was drawn through the Tenax and Tenax/charcoal
tubes at a nominal flow rate of 0.5 L/min for 40 min, for a
total sample of 20 L. The sample tube sets were submitted
to quality control (QC) contamination checks prior to use and
were stored refrigerated at 1*0 in Teflon bags both prior to and
after use. Daily field  blanks were performed, and all samples
were analyzed within 30 days.
  The VOST samples were analyzed by GC/MS on a purge and
trap system devoted to VOST sample analyses. Method 5040 of
                       SW-846 best represents the procedure used for sample analy-
                       ses (25). The unknowns were identified using mass spectral
                       library searches as well as investigator interpretation. Quan-
                       tification of the identified unknowns was based on the toluene
                       response factor obtained during initial calibration.
                          Prior to calibration or  the analysis of samples, the MS
                       was tuned with perfluorotributylamine  (PPTBA) to linearize
                       the working range of mass  units (45-420 amu). Following
                       initial linearization,  a multipoint calibration using toluene

-------
4818
TIRES, OPEN BURNING
was performed. This calibration was used to quantify the QC
samples analyzed at the beginning and end of each sample
period as well as after every third sample. The QC sample
contained  known concentrations  of toluene and bromofluo-
robenzene (8FB). The BFB was used to confirm ion abundance
criteria, thereby verifying mass linearity of the instrument.
   As stated earlier, the Tenax tube samples were analyzed on
a dedicated purge and trap GC/MS system. The samples were
desorbed in a clamshell heater maintained at 190*0 using a he-
lium carrier at a nominal flow rate of 25 mLAnin onto a Tenax
trap at room temperature for 10 min. At this point, the trap
was heated rapidly to 225'C, and the carrier was directed onto
a 30 m DB-624 megabore column. The carrier flow at this con-
dition was nominally 5.6 mL/min. The oven temperature was
maintained at 30*C for 5 min; then a temperature ramp was
invoked at 5'C/min until reaching 160'C, when  the tempera-
ture was held for 15 more minutes. As the sample constituents
eluted from the column, they passed through a jet separator
before being introduced into the MS. A spectral sweep from 45
to 420 amu was performed each second. Data were acquired
and stored by computer. Peak areas were integrated by man-
ually establishing baseline and integration limits. Unknowns
were identified using the system mass  spectral library, em-
ploying both the forward and reverse searching capabilities.
Compound boiling point was also used to help identify the
unknown.

   Semivolatile Organics. Semivolatile  organics were collected
on both the participate filters and the XAD-2 organic sor-
bent. The system used for this effort is shown in Figure  3. The
Teflon-coated fiber filter used for particulate collection was
desiccated, tared, and placed in aluminum foil and a resealable
plastic bag prior to use. After sample collection,  the sam-
ples were wrapped and stored refrigerated at  1'C until being
desiccated, weighed, and extracted. The filter housing sys-
tem was located immediately upstream of the XAD-2 canister.
Cleaned and tested XAD-2 resin was placed in the  canisters,
sealed in Teflon bags, and stored refrigerated  prior to use.
After sampling, the canisters were reseated in the Teflon
bags  and stored refrigerated until being extracted less than
14 days later. The gaseous sample was collected at an average
flow rate of 2-2.5 cfm for ~3 h. During the  CHUNK condi-
tion, the participate filters became loaded to the point that
replacements were required. This, however, did not affect the
completeness of the samples because the sampling periods cor-
responds to a known mass displacement.
   Organics were  retrieved  from  the  collection media by
Soxhlet extraction using dichloromethane. The XAD-2 was
extracted separately from the participate fraction. Following a
24-h extraction, the samples were concentrated to 10 mL using
a three-ball Snyder column system. All organic concentrates
were stored refrigerated until needed.
   Both the particulate extracts and the  XAD-2 extracts were
analyzed for total chromatographable organics (TCO) (organic
compounds with boiling points between 100 and 300'C) and
GRAV (organic  compounds with boiling points  greater than
300*C). The TCO analysis was done by GC/FID (26). A multi-
point calibration was conducted using a mixture of alkanes.
The 7,10,12,14. and 17 carbon alkanes were used to quantify
and identify the temperature window  for analysis. All peaks
with retention times falling between but not including the C7
and Civ retention times were quantified. A response  factor
                                                calculated from the average of Cw, Cu, and CM areas was
                                                used for quantification. The analysis was performed using a
                                                30 m DB-5 megabore column with a flow rate of 8 ml/min.
                                                The 2-5 mL injection was made with  the oven temperature
                                                held at 40*C for 3 min, then ramped to 250'C at 20'C/min and
                                                held for 15 min after reaching final temperature.
                                                   The GRAV fraction was measured by direct weighing. Alu-
                                                minum weigh boats were desiccated, tared, and then filled with
                                                0.5 mL of the organic extract and allowed to evaporate. After
                                                evaporation, the boats were again desiccated and weighed. Or-
                                                ganic compounds with boiling points greater than 300*C repre-
                                                sent the net gain. The analysis was performed in duplicate and
                                                included an audit sample.
                                                   Unknown organics were identified using a GC/MS system.
                                                For liquid samples, a  Hewlett Packard gas chromatograph/
                                                mass selective detector  (GC/MSD) system configured  for
                                                capillary columns was used. Split injections of 1-2 fiL at a
                                                ratio  of 100:1 were  introduced onto a  30-m SPB-5 capillary
                                                column. An initial oven temperature of 40'C was maintained
                                                for 5  min before ramping the temperature to 250*C at 5"C/
                                                min. The final temperature was held for 15 min. Compounds
                                                were  identified using  library spectral matching. The Wiley
                                                Library was  used during spectral searching and matching.
                                                Again, boiling points were used in assisting investigator inter-
                                                pretation in determining compound identity. In several cases,
                                                the mass spectra of known standards  were  used to confirm
                                                identifications.
                                                   Unknowns  were  quantified through  comparison  of the
                                                GC/MSD runs with injections  run  under similar conditions
                                                using a GC/FID system. The SPB-5 column is virtually iden-
                                                tical to the DB-5 column used, the only real differences being
                                                in the manufacturer, stationary phase thickness, and column
                                                ID. The injections were made on the same system as was used
                                                for the TCO analyses. The only change made was to alter the
                                                temperature ramp to match that of the GC/MSD runs. Several
                                                standard mixes containing compounds identified by GC/MS
                                                were prepared and run on the GC/FID system. The retention
                                                times of these standards were used as markers to relate the
                                                MS runs with the FID runs. The elution order of the MS runs
                                                was assumed to be identical to that of the FID runs. A linear
                                                relationship between  the retention times of the runs was
                                                determined, and, along with comparison of peak magnitude,
                                                compound  identifications were assigned  to the FID runs.
                                                Individual  peaks  were quantified with  the same response
                                                factor used for TCO quantifications.
                                                   A portion of the liquid concentrates was also analyzed for
                                                polycyclic aromatic hydrocarbons (PAHs). This analysis was
                                                subcontracted and used EPA Method 610 as the referenced pro-
                                                cedure (27).

                                                   Airborne  Metals Particulate. Particulate matter  was col-
                                                lected using a separate sampling system in order to character-
                                                ize airborne metals emissions.  A gaseous sample was drawn
                                                through a 142-mm quartz-fiber filter under vacuum at an av-
                                                erage flow rate of 2-2.5 cfm for ~3 h. During the CHUNK
                                                condition, two particulate filters were required. The quartz fil-
                                                ters used were desiccated and tared, then placed in aluminum
                                                foil and a resealable plastic bag prior to  use. Following sam-
                                                ple collection, the samples were refrigerated until they were
                                                again desiccated and re-weighed. Ultimately, the samples were
                                                delivered to an outside analytical laboratory for metals quan-
                                                tification. Specific metals common in tire-ash  residue were

-------
chosen for quantification (28). The samples were analyzed us-
ing inductively coupled argon plasma  (ICAP) methodology
(29).

   PMio Participate Collection. An Andersen medium volume
ambient particulate sampler was used to collect particulate of
10 /tim in diameter or less. The sampler is designed so that,
when a flow of 4 cfm is maintained on the system, particulate
of 10 /zm in diameter or less only is collected on the filter. The
110-nun fiber filters were desiccated and tared, then placed in
aluminum foil and a resealable plastic bag prior to use. Follow-
ing sampling, the  filters were desiccated and weighed to deter-
mine total amount collected.

EPA Bioassay Tests

The samples used in the EPA bioassay tests were  the same
samples generated during the earlier EPA emissions character-
ization tests. Organics were recovered from filters and XAD-2
by 24-h Soxhlet extractions using dichloromethane (DCM), and
the DCM-ex tractable mass was determined gravimetrically.
The particulate (filter) organics were fractionated by HPLC as
follows: 500 fi.g of DCM extractable mass from each  burn con-
dition was injected into a Varian 5560LC containing an Econo-
sphere silica 3-fim column (4.6 X 150 mm). The step gradient
started at 100% n-pentane, which was held for 15 min, followed
by a 25-min gradient to 100% DCM, which was held for 15 min,
followed by a 5-min gradient to 100% methanol, which was held
for an additional  5 min. All gradients  were linear,  all flows
were 1 mL/min, and 1 fraction (1 mL) was collected per minute,
resulting in 60 fractions/sample.

Ontario Field Samples

Approximately 1  week after  the  Ontario tire fire began,
a PM]0 ambient  sampler was placed approximately 250 m
from the southern perimeter of the tire fire burn directly in
the plume. Respirable  particulate material was collected at
1.13 m3/min on Teflon impregnated glass fiber (TIGF) filters
(Pallflex, No. T60A20, 2Q.3 x 25.4 cm)  for 20 h. Filters were
changed at approximately 10-h intervals, but flow rates were
not maintained at the initial rate because the filters became
clogged by the fine particulate matter in the smoke.  Although
no adjustment was made for the reduction in flow rate over the
sampling period,  the total air flow over the sampling period
was estimated to be 1360 m3.
   The organics were recovered from the filters by Soxhlet ex-
traction for  16 h with DCM followed by 16 h with methanol.
The extracts were pooled, the mass was determined gravimet-
rically, and the organics were fractionated on neutral alumina
(1 X 30 cm) to yield four fractions of increasing polarity. The
fractions and the  solvents used to elute  them were Al (60 mL
of hexane), A23 (50 mL of benzene followed by 70 mL of chlo-
roform with 1% ethanol), A45 (50 mL of methanol followed by
50 mL of methanol/water at 4:1), and A6 (50 mL of water). The
aliphatics that were present in fraction  A23 were removed by
chromatography on Sephadex LH20 (3.5 X 30 cm) with hex-
ane:methanoi:dichloromethane at 6:4:3 to produce an aromatic
fraction designated A23LH20 that was suitable for GC analysis.
   Normal-phase  HPLC  was  performed on  the A23LH20
fraction using a 1.5-cm amino precolumn (4.6 mm ID, Brown-
lee  Laboratories,  Santa  Clara,  California), and  a  10-fj.m
Whatman Partasil M9 PAC analytical  column (25  x 9.4 cm
ID, Whatman, Clifton, New Jersey).  A 100-/*L sample loop
                            TIRES, OPEN BURNING      4819

and mobile-phase flow rate of 4.2 mL/min were used to elute
eight subfractions designated as NO through N7, which were
collected as follows (fraction, time to collect): NO, 7.5 min; Nl,
24 min; N2,30 min; N3,34 min; N4, 39 min; N5,45 min; N6,
50 min; and N7,65 min. The following linear saivent-gradient
program was used for elution (elapsed time, composition of
mobile phase): initial,  100% hexane;  5 min,  100% hexane;
10 min, hexane:DCM at 99:1; 15 min, hexane:DCM at 95:5;
50 min, 100% DCM; 65 min, 100% ethanol.
   Chemical analyses of the whole extract and of the Nl sub-
fraction were performed on a Hewlett Packard 5890 Series II
GC equipped with a splitless injector and a 25-m, DB-5 fused
silica capillary column (0.25 mm ID, J&W Scientific, Folsom,
California). A Hewlett Packard 5971 mass selective detector
was used for GC/MS studies. Hewlett Packard ChemStation
MS  software was  used to control the  instruments  and to
acquire and process the data. Samples (1  jiL) dissolved in
glass-distilled toluene were injected and analyzed using the
following temperature  program: 50 to  160'C at  20'C/min;
160  to 290*C at 3*C/min; and hold at 290'C for 10 min. The
addition of an internal standard (9,10-dimethylanthracene)
was used to quantitate the concentrations of compounds in the
extracts.

Bioassay Methods
Although the Salmonella mutagenicity assay is performed in
a variety of ways and with different strains of the Salmonella
bacteria, the assay is generally performed as follows. The or-
ganic extract of the combustion emission is solvent exchanged
into dimethyl sulfoxide (DMSO), which is a relatively nontoxic
solvent. Then, 100 fiL of a suspension of the bacteria ere added
to molten agar (at 45*C) along with various amounts of the or-
ganic extract (1-100 /j.g in a volume no greater than 100 /*L),
and  the mixture is poured onto the surface of an agar petri
plate. After the plates are incubated at 37*C for 2-3 days, bac-
terial cells that contain an appropriate mutation in the appro-
priate gene will grow into a visible cluster (colony) of cells on
the surface of the plate. (Bacterial cells that do not incur an
appropriate mutation will not grow into a visible colony.) The
resulting colonies are then counted, and a dose-response curve
is generated. The mutagenic potency of the agent is calculated
from the slope of the linear portion of the dose-response curve
and is expressed as mutants/microgram or, as described subse-
quently, revertants/microgram.
   Normal strains of Salmonella bacteria can grow on nothing
more than a carbon source (usually a sugar such as dextrose)
and  a nitrogen source (usually a salt such as sodium ammo-
nium phosphate). For various reasons, the bacterial strains
used in the assay contain certain mutations, and one of these
mutations prevents the cell  from synthesizing an essential
amino acid (histidine) from just sugar and nitrogen in the agar
medium. Thus, the strains of bacteria used in the assay cannot
grow into visible colonies on the medium used in  the assay,
which contains just sugar and  a nitrogen  source. However,
when a particular type of mutation occurs within the already
mutated histidine  gene,  this second  mutation can  reverse
the effect of the first mutation by correcting (completely or
partially) the defect in the histidine gene and permitting the
cells to grow in the absence of exogenous histidine. Such cells
are considered to have been mutated  back to normal. Thus,
the assay is called a reverse-mutation assay, and the newly

-------
4820
TIRES, OPEN BURNING
mutant colonies are called revertants, implying that they have
been reverted to normal.
   Another critical feature of the assay is the incorporation
of an enzyme preparation (usually consisting of a crude ho-
mogenate of rat liver) into the top agar along with the organic
extract and the cells. This enzyme preparation is called S9 mix,
which derives from the fact that it is the supernatant from a
9,000 X g centrifugation of a homogenate of rat liver. Studies
during the past 25 years have indicated that many potential
mutagens and carcinogens in the environment, such as PAHs,
are not mutagenic perse, ie, they do not bind covalently to DNA
in their native form. Instead, they are converted (metabolized)
to electrophilic forms by enzymes in the body, and it is the elec-
trophilic forms that can bind to DNA and lead to mutations
and cancer. Bacteria are generally lacking the key enzymes in-
volved in this metabolism. Thus, the inclusion of 89 mix par-
tially replicates mammalian metabolism in the petri dish and
permits the detection of a wide range of environmental muta-
gens and carcinogens that would otherwise not be detected as
 mutagenic in the absence of the enzyme preparation.
    Laboratory Samples.  All of the organic extracts were solvent
 exchanged into DMSO for bioassay (19). All of the crude ex-
 tracts from the laboratory tire burns were evaluated for muta-
 genicity by using the plate-incorporation assay in strain TA98
 of Salmonella with and without Aroclor 1254-induced Sprague-
 Dawley male rat liver S9 as described  elsewhere (30). A
 composite sample of all of the laboratory tire burns was eval-
 uated in strains TA98 (+/-  S9), TA98NR (-S9), and TA98/
 1,8-DNP6  (-S9). These strains were  used  to estimate the
 contribution made by nitroaromatics to the total mutagenic ac-
 tivity observed in the emissions in strain TA98 (-S9). Strains
 TA98NR and TA98/1,8-DNP6 were used  in conjunction with
 the standard tester strain TA98  because  reduced mutagenic
 activity in the first two strains relative to TA98 indicates the
 presence of nitroaromatics.  This reduced mutagenic activ-
 ity occurs because strains TA98NR  and TA98/1,8-DNP6 are
 deficient in  the nitroreductase and  transacetylase enzymes,
 respectively, that are  necessary for  the conversion of mono-
                                                 and dinitroaromatics, respectively, to mutagenic electrophUic
                                                 arylhydroxylamines (31).
                                                   Because most nitroaromatics are activated to mutagens by
                                                 bacterial enzymes, no exogenous metabolic activation in the
                                                 form of rat liver S9 is required for these strains. In addition,
                                                 S9 partially inactivates nitroaromatic compounds, complicat-
                                                 ing the interpretation of the data obtained in the presence of
                                                 S9 (32). Regressions over the linear part of the dose-response
                                                 curves were calculated to determine-mutagenic potencies (re-
                                                 vertants/microgram) or mutagenic concentrations (revertants/
                                                 cubic meter). A microsuspension assay was used as described
                                                 previously (33) to evaluate the mutagenicity of the HPLC frac-
                                                 tions (16,19,34).

                                                    Ontario Field Samples. Preliminary experiments with the
                                                 extracts from the field tire burn indicated that the extracts
                                                 were more mutagenic in strain YG1024, which is similar to
                                                 TA98  but has enhanced  0-acetyltransferase activity (35),
                                                 than they were in strain TA98. Thus, all further studies with
                                                 the field tire burn extracts were performed using the plate-
                                                 incorporation assay with strain YG1024, which was kindly
                                                 provided by M. Watanabe, National Institute of Hygenic Sci-
                                                 ences, Tokyo, Japan.

                                                 RESULTS

                                                 Laboratory Data: Burn Rate  Results
                                                 As stated earlier, the size of tire material was varied to alter
                                                 the combustion conditions and to gain insight into the mecha-
                                                 nisms governing burn rate. Rates were calculated by dividing
                                                 the amount of tire material burned by the length of the burn
                                                 and then normalizing to a mass per hour basis. The data show
                                                 that the CHUNK condition produced a higher burn rate than
                                                 the SHRED condition. Almost double the amount of CHUNK
                                                 material was combusted during similar periods.
                                                     Figure 4 is a plot of the burn rate versus the elapsed time
                                                 for all tests. All plots show a high initial burn rate, but as time
                                                 elapsed, the burn rate diminished. In both of the SHRED runs,
                                                 after more than midway through the test period, the burn  pit
                                                 had to be agitated to sustain combustion. This agitation would
                                        i,
                                        3C
                                        of
                                        IS
                                         DO
                                         C
                                         3
                                         .O
 Figure 4. Burn rate vs time for all tests:
      (+) - CHUNK Day 1; (O) CHUNK
       Day 2; (•) SHRED Day 1; and (^)
                        SHRED Day 2.
                                                                100          150
                                                                 Elapsed time, min
200
250

-------
                                                                                     TIRES, OPEN BURNING     4821
account for the visible increase in burn rates occurring at that
point.
   Under both conditions, an initial high burn rate was ob-
served to gradually level off until & steady rate was achieved.
Examination of the tire material suggested a possible explana-
tion. Along the tread portion of the tires, a much thicker layer
of rubber was found in relation  to the sidewall. In addition,
the rubber of the sidewall portion seemed impregnated into the
belt material, yet the rubber in the tread surface contained no
belt material. It may be that the tread material is consumed
first while the belt containing rubber may be more difficult to
burn and more uniform in its combustion.

Laboratory Data: CEM Data

Combustion • gas  products  were  monitored continuously
throughout the test period. Figure 5 illustrates the data from
the CO, S02, and THC analyzers for the four tests. The O2
and COs CEM data did not deviate significantly from ambient
concentrations (O2 never dropped below 20%  and C02 never
rose above 0.4%), so are not included in these plots. Comparing
these plots with the respective plots pertaining to burn rate
shows a relationship between high emissions of CO, SO?, and
THC at high burn rates.

Laboratory Data: VOC Data

Identification of unknown organics using the MS proved to be
highly successful. Table 1 lists the compounds identified, their
average  gaseous  concentrations, and estimated emissions
at the CHUNK and SHRED conditions.  These values are
estimates and are  calculated using the MS response to toluene
from the VOST samples collected during testing. The majority
are aliphatic, olefinic, or acetylenic substituted anomalies. The
 s
 §
 8
500
400
300
200
100
500
400
300
200
100
500
400
300
200
100
500
400
300
200
100
                         CHUNK day 1
                         CHUNK day 2
SHRED day 1
                 Agitation
                 point
                50       100      150
                       Elapsed time, min
                 200     250
Figure 5. Concentration of combustion gas products over time.
predominant formation of aromatic hydrocarbons is likely due
to the high thermodynamic stability of aromatic structures.
A representative of each compound class is found  in mono-
through polysubstituted aromatic hydrocarbons. Cyclic alka-
ncs, alkenes, and dienes were also present. It is not surprising
to identify butadiene in  the samples because  it is a major
constituent of the tire fabrication process. A halogenated
compound,  trichlorofluoromethane, was  also  identified in
several  of  the  collected samples.  This  chlorofluorocarbon
(CFG), also known as CFC-11, was probably emitted by the
air conditioners used to dilute the air in the burn hut. Several
sulfonated compounds were identified in  the samples. Thio-
phene and substituted thiophenes were isolated. Nitrogenated
hydrocarbons were found. Isocyanobenzene and benzodiazine
were isolated in multiple samples.
   The data presented in Table 1 represent an averaging of
the three sets of VOST samples t^ken at each run condition,
each taken at different periods during the burn. The data did
not reveal consistent trends in either the types or amounts
of emissions under varied burn rates. Benzene is emitted in
large quantities under all conditions. Average gaseous concen-
trations increase with increased burn rate, but this is true with
the majority of the compounds presented. It is  interesting to
note, however, that, as the burn rate decreased, the mass of
specific compounds emitted tended  to increase with respect
to the amount of tire material combusted. It may be that dur-
ing the latter portion of the burn period, the remaining rubber
in the tire material was bound with the cord material and be-
came difficult to burn. In this lower temperature regime, the
rubber continued to be pyrolyzed, but less was combusted while
the volatiles reacted to form the types of compounds identified.
   The estimated emissions presented are based on several
variables. They were obtained  by assuming that the dilution
air added to the burn hut was at a constant volume and that
the amount of air added equaled the amount exiting the hut.
It was also  assumed that  the gas  mixture collected in  the
sample  duct was well mixed and representative of the  gas
mixture found throughout the burn hut. The average gaseous
concentration or average concentration of the sample over a
given period was determined by dividing the total collected
amount (obtained by GC/MS analysis) by the volume of sample
collected. This value was then multiplied by the amount of
air added to the burn hut in 1 h. This value is the amount of
a specific compound emitted on  an hourly basis, and it  was
then divided by the burn rate determined for that period.  The
burn rate, as outlined earlier, was determined by dividing the
amount of tire material combusted in a specific period by  that
amount of time (in minutes) and normalizing it to an hourly
basis. The resultant value is  an estimate of the amount of
compound emitted at a specific burn rate.  Finally, the total
for the day's runs was derived by averaging the results for
the three VOST tubes based on total sample volume. The two
CHUNK runs were then averaged together, as were the two
SHRED runs.
                                                     Laboratory Data: Semivolatile Organic Compound Data

                                                     As with the volatile organic analyses, the MS analysis of the
                                                     XAD-2 extracts identified the same types of compounds. Ta-
                                                     ble 2 lists the 34 compounds identified and indicates that sub-
                                                     stituted mono- and polyaromatics were again the predominant

-------
4822
         TIRES, OPEN BURNING
Table 1. Quantitatlon and Emission Summary of Compounds Identified in VOST Samples*
Compound Identified
Benzaldehyde
Benzene
Benzodiazine
Benzofuran
Benzothiophene
Butadiene
Cyclopentadiene
Dihydroindene
Dimethyl benzene
Dimethyl hexadiene
Dimethylmethylpropyl benzene
Dimethyldihydro indene
Ethenyl benzene
Ethenyl cyclohexene
Ethenyldimethyl benzene
Ethenylmethyl benzene
Ethenyldimethyl cyclohexene
Ethenylmethyl benzene
Ethyl benzene
Ethylmethyl benzene
Ethynyl benzene
Ethynylmethyl benzene
Heptadiene
Isocyano benzene
Limonene
Methyl benzene
Methyl cyclohexene
Methyl hexadiene
Methyl indene
Methyl naphthalene
Methyl thiophene
Methylethenyl benzene
Methylmethylethenyl benzene
Methylmethylethyl benzene
Methylmethylethyl cyclohexene
Methylpropyl benzene
Methylene indene
Methylethyl benzene
Naphthalene
Pentadiene
Phenol
Propyl benzene
Tetramethyl benzene
Thiophene
Trichlorofluoromethane
Trimethyl benzene
Totals

Concentration,
mg/m3
0.260
1.910
0.017
0.049
0.014
0.152
0.081
0.013
0.413
0.008
ND
0.007
0.678
0.006
0.014
0.016
ND
0.129
0.182
0.120
0.322
O.S62
0.009
0.341
0.011
0.976
0.005
0.021
0.138
0.287
0.006
0.027
0.046
0.041
ND
ND .
0.038
0.045
1.285
0.077
0.002
0.026
ND
0.023
0.158
0.022
8.534
CHUNK
Estimated Emissions,
ing/kg tire
299.2
2,156.3
13.7
25.1
26.3
308.4
48.6
40.6
779.7
28.3
ND
22.0
941.8
26.2
7.2
14.1
ND
221.6
460.8
334.5
190.0
530.6
25.4
347.4
27.5
1,606.4
21.1
71.3
316.4
311.5
5.5
55.7
98.0
111.4
ND
ND
48.5
134.9
1,130.0
163.9
0.5
72.4
ND
54.6
57.6
46.9
11.182.0

Concentration,
mg/m3
0.215
1.401
0.014
ND
0.011
0.096
ND
0.021
0.629
0.049
0.008
0.008
0.395
0.060
0.014
0.014
0.193
0.028
0.164
0.262
0.110
0.226
0.028
0.191
0.513
0.714
0.023
0.068
0.087
0.135
. 0.007
0.045
0.373
0.165
0.086
0.020
0.022
0.092
0.607
0.680
0.016
0.046
0.130
0.021
ND
0.042
8.029
SHRED
Estimated Emissions,
rag/kg tire
329.6
2,204.8
17.4
ND
14.7
160.8
ND
42.8
1,078.2
90.9
14.9
17.7
611.4
107.6
23.7
19.5
350.4
40.9
295.1
475.8
131.5
258.7
51.4
289.7
892.5
1,129.1
40.1
127.2
140.1
196.6
12.6
76.6
683.5
283.5
170.0
41.6
34.4
169.4
823.6
1,162.6
14.3
84.2
255.7
27.S
ND
74.9
13,067.8
• Concentrations determined using system response to toluene. These data are averaged over six sets of VOST tubes per day taken over 2 days. ND is none detected.
products of incomplete combustion (PICs). The data in Table 2
are an average of three samples taken over the entire course of
the day's run. The quantities found in each individual sample
roughly parallel the burning rate; ie, the early samples had
a higher loading and the later samples had a lower loading
of semivolatile organics. Many of the compounds collected and
identified in the volatile organic samples were also found in the
XAD-2 extracts. Table 2 shows that similar ambient-loading
and estimated emission values were realized from the XAD-
2 extracts. The estimated emissions for many compounds, but
not for all, increased with decreasing burn rate. An exception
to this, for example, is naphthalene, in which emissions were
greater at higher burn rates, as evidenced during the CHUNK
condition. In addition,  the  average gaseous concentration is
also greater in this situation. This finding contrasts with the
trends observed in the VOST samples.
  The particulate  filters  located upstream of the XAD-2
canisters contained  considerable quantities of organics. The

-------
 Table 2. Quantitation and Emission Summary of Compounds Identified in XAD-2 Extracts*
                                              CHUNK
                                                                                         TIRES, OPEN BURNING      4823
                                                               SHRED
 Compound Identified
Average Concentration,
       mg/m3
Estimated Emissions,
     ing/kg tire
Average Concentration,
       mg/m3
 ° Concentrations determined using system response to total ehromatographable organic calibration mix. ND is none detected.
Estimated Emissions,
     mg/kg tire
1 -Methyl naphthalene
1 , 1 '-Bipheny Imethyl
Itf-Fluorene
2-Methyl naphthalene
Acenaphthylene
Benzaldehyde
Benzisothiazole
Benzo(6)thiophene
Biphenyl
Cyanobenzene
Dimethyl benzene
Dimethyl naphthalene
Ethyl benzene
Ethyldtmethyl benzene
Ethynyl benzene
Hexahydro azepinone
Indene
Isocyano naphthalene
Limonene
Methyl benzaldehyde
Methyl benzene
Methyl indene
Methylmethylethyl benzene
Methylethyl benzene
Naphthalene
Phenanthrene
Phenol
Propenyl naphthalene
Propenylmethyl benzene
Propyl benzene
Styrene
Tetramethyl benzene
Trimethyi benzene
Trimethyl naphthalene
Totals
0.292
0.013
0.187
0.314
0.580
0.218
ND
0.050
0.186
0.199
0.254
0.034
0.181
ND
0.254
0.062
0.462
0.011
0.047
ND
1.105
0.093
0.107
0.040
1.578
0.173
0.330
0.027
ND
ND
0.605
ND
ND
ND
7.593
330.7
11.1
210.3
350.7
633.8
244.1
ND
44.2
209.5
223,7
305.0
41.1
205.2
ND
275.8
75.1
5C3.4
9.4
56.1
ND
1,212.2
111.8
127.9
48.3
1,697.9
183.7
365.9
23.5
ND
ND
659.9
ND
209.4
ND
8,369.7
0.133
ND
0.183
0.255
0.318
0.180
0.094
ND
0.193
0.300
0.544
0.096
0.197
0.158
0.112
0.445
0.201
ND
1.361
0.047
0.816
0.234
0.821
0.133
0.671
0.119
0.412
ND
0.282
0.127
0.380
0.049
0.446
0.185
9.492
227.6
ND
308.4
429.2
531,1
333.9
173.9
ND
330.1
516.8
935.1
178.1
337.6
272.4
187.4
748.5
339.2
ND
2,345.5
86.6
1,390.1
400.7
1,426.1
229.1
1,130.7
187.0
700.2
ND
523.6
219.6
645.5
91.9
751.4
315.8
16,293.1
Table 3. Organic Emission Summary

Average Gaseous Concentrations, mg/m



Estimated Emissions, mg/kg tire



Range

TCO
GRAV
Total

TCO
GRAV
Total
CHUNK

3,514.6
4.048.0
7,562.6

9,792.0
11,223.5
21,015.5
SHRED

8,473.0
4,151.9
12,624.9

31,686.0
14,888.0
46,574.0
majority of compounds found in the particulate have boiling
points exceeding  300'C. Table 3 summarizes semivolatile
organic emission data from all four test conditions. The total
organics  from  respective boiling-point-based  analyses are
presented for each component of the sampling media. The
estimated emissions data presented show that from 12 to 50 g
of semivolatile organics can be emitted for every 1 kg of tire
burned. It appears that, as burn rate decreased, the amount of
                               organics emitted, particularly in the total ehromatographable
                               organic (TCO) range, increased. There did not seem to be any
                               significant variation in GRAV range organic emissions when
                               related to bum rate.
                                 The results of the  HPLC  PAH analyses are presented
                               in Table 4. The XAD-2 and particulate filter extracts were
                               analyzed separately. The particulate filter extracts contained
                               compounds mostly in the GRAV range, and most semivolatile

-------
4824     TIRES, OPEN BURNING
           Table 4. PAH Quantitation and Emission Summary
CHUNK
Compound
Acenaphthene
Acenaphthylene
Anthracene
Ben2(a)anthracene
Benzo(o)pyrene
Benzo(6 )f luoranthene
Benzo(g,/t,i)perylene
Benzol A )f luoranthene
Chrysene
Dibenzta, h (anthracene
Fluoranthene
Fluorene
Indeno(l,2,3-cd)pyrene
Naphthalene
Phenanthrene
Pyrene
Totals
Average
Concentration,
mg/ma
0.282
0.802
0.053
0.076
0.080
0.064
0.060
0.069
0.068
0.001
0.324
0.243
0.049
0.786
0.225
0.030
3.212
Estimated
Emissions, mg/kg
tire
290.3
861.3
56.3
82.2
84.8
69.4
66.0
74.3
70.8
1.1
338.7
260.5
51.6
815.9
237.5
33.8
3,394.5
SHRED
Average
Concentration,
mg/mg8
1.404
0.334
0.029
0.062
0.068
0.053
0.095
0.059
0.056
ND
0.273
0.112
0.051
0.289
0.149
0.090
3.124
Estimated
Emissions, mg/kg
tire
2445.7
561.8
49.6
102.4
113.9
88.4
159.4
99.4
91.6
ND
458.0
186.8
85.5
486.0
252.5
151.7
5,332.7
 compounds were not able to be identified by GC/MS. However,
 HPLC was used to identify PAHs from both the XAD-2 frac-
 tion and the particulate filter fraction. The 16 PAHs include
 several compounds known to be carcinogenic. In particular, the
 presence of benzo(a)pyrene (BAP) is of major importance. It is a
 highly scrutinized compound when evaluating combustion pro-
 cesses. Although no trend in concentration related to burn rate
 exists, the magnitude of the emissions warrants concern. Dis-
 crepancies in reported PAH values are likely a result of the an-
 alytical methods used. The HPLC method is specific to PAHs
 and, due to its specificity, is far less susceptible to interferences
 from other compounds present. The FID quantitative method
 used responds to hydrocarbons in general and is more suscepti-
 ble to chromatographic interferences/coelution problems, par-
 ticularly given the complexity of the sample matrix.
   The HPLC PAH analysis accounted for only roughly 10% of
 the GRAY range organics. It was not possible to identify the
 remaining portion because of equipment limitations. The GCV
 MS system used was not set up for high temperature applica-
 tions, so this area remains unexplored. It may be possible that
 carbon black, a major constituent of tire material, may exist
 as sub 0.45  fim particles that passed through the filter dur-
 ing cleanup of the particulate extraction. It may also be possi-
 ble that some of the discrepancy between the PAH and GRAY
 results may be due to suspected low sample recoveries for the
 PAH analysis.

 Laboratory Data: Particulate Data

 The particulate was collected using three separate systems.
 Particulate was captured with the semivolatile organic sys-
 tem, with the airborne metals particulate collection system,
 and with a medium volume (4 ftVmin (0.11 m'/min)) ambient
 PMio sampler located in the burn hut. Table 5 summarizes the
 particulate loading values of these three systems for the four
 test conditions. For total average gaseous concentration there
seems to be  good agreement between the organic particulate
and the metals particulate systems during each test condi-
tion. Moreover, average gaseous concentration increased with
increased burn rate. As the burn rate decreased, the percent
of organics extracted increased. This finding is important be-
cause, although under reduced average gaseous concentration,
the amount of organic material that the particulate contains is
greater. Comparing burn rate to estimated particulate emis-
sions shows  that the emission  rate of particulate decreased
with lower burn rates, although nearly 100 g of particulate was
emitted for every 1 kg of tire combusted.
  The PMW sampler was used to assess the amount of particu-
late found in the 10 fim or less range. Owing to the nature of
the sampler and the application to this study, several problems
were encountered during data collection. The ambient sampler
was designed to operate constantly at 4 cfm (0.113 m3/min).
This relatively high flow rate was difficult to maintain because
of the rapid loading of the particulate filter. As the loading in-
creased, the flow rate through the system decreased until the
required flow rate could no longer be obtained. The 4-cfm flow
rate was required to maintain the specific cutoff point for par-
ticulate sizing. As the flow rate decreased, the size of particu-
late reaching the filter increased; therefore, the data presented
may not be valid because the required flow rate was not main-
tained, although sampling still  continued despite inadequate
flow.
Laboratory Data: Airborne Metals Data

A separate particulate collection system was operated to verify
and quantify the presence of metals collected from ambient
emissions. Seventeen target metals reportedly found in tires
were isolated.  The list was compiled from information on
combusted-tire ash residues. Table 6 shows the results of the
metal analyses. The results from the  method blank are also
included  to demonstrate  the marginal differences between
the blank and collected samples. Many of the analyses are
at or near instrument detection levels. The only significant
differences between the blank  and sample were found  with
lead and  zinc. The lead results  are extremely close to the in-
strument detection level of the  element. The major difference

-------
                                                                                       TIRES, OPEN BURNING      4825
           Table 5. Participate Collection Summary
                                        CHUNK
                       SHRED
Sample
Organic
Particulate
Filter
Metal
Particulate
Filter
PM,n Filter-
Average Cone.,
mg/m3
93
111.55
444.14
Estimated
Emissions,
mg/kg tire
97,100
105,000
113,500
Extractable Average Cone.,
Organic, % mg/m3
10.6 43.75
37.9
92.85
Estimated
Emissions,
mg/kg tire
73,400
64,500
149,000
Extractable
Organic,'*
19.65
           • The PM,0 sampling niter became heavily loaded during the initial part of each run. The results are biased high, due to higher
           burning rates that occurred during this portion of the run.
           Table 6. Airborne Particulate Metals Analysis Summary

                                                                           Emissions
                                                              CHUNK
                              SHRED
Filter Totals, ing
Metal
Aluminum
Antimony
Arsenic
Barium
Calcium
Chromium
Copper
Iron
Lead
Magnesium
Nickel
Selenium
Sodium
Titanium
Vanadium
Zinc
Blank
0.020
<0.02
<0.0005
<0.01
0.030
0.010
<0.002
0.060
<0.0005
0.004
0.010
<0.0005
0.020
<0.05
<0.05
0.010
CHUNK
0.019
<0.02
<0.0005
<0.01
0.050
0.010
<0.002
0.060
0.002
0.007
0.011
<0.0005
0.053
<0.05
<0.05
0.206
SHRED Cone., mg/m3
0.020
<0.02
<0.0005
<0.01
0.040
9.014
<0.002
0.065
0.001
0.007
0.009
<0.0005
0.050
<0.05
<0.05
0.215


0.0079



0.0004
0.0012


0.0084


0.0409
Estimated
Emissions,
mg/kg tire


8.54



0.47
1.26


9.51


31.17
Cone., mg/m3


0.0028



0.0001
0.0005


0.0035


0.0146
Estimated
Emissions,
mg/kg tire


4.80



0.10
0.75


5.80


24.35
 was found in the zinc analyses. Many of the estimates are
 based on the detection levels themselves and are presented as
 less-than quantities. The zinc data suggest that both average
 gaseous concentration and estimated emissions increased with
 inci. '-^sed burn rates.

 Bioassay Data

   Laboratory  Data; Mutagenic Potencies.  Table 7 shows the
 mutagenic potencies (revertants per rnicrogram) of the DCM-
 extractable organics based on the slopes of the mutagenicity
 dose-response curves (19). The mutagenic potency (revertants
 per microgram) of the semivolatile organics (XAD) was similar
 for CHUNK and SHRED tires: ie, approximately 1-2 rev/fig.
 The exception was Day 2 of the SHRED XAD sample, which
 was considerably more potent, perhaps due to the agitation of
 the burn pit during this burn in order to sustain combustion.
The mutagenic potency (revertants per microgram) of the par-
ticulate organics (filter) was generally greater than that of the
semivolatile organics, ranging from approximately 2 to 12 rev/
fj.g. In the presence of 89, the participate organics from the
CHUNK tires exhibited greater mutagenic potency than those
from the SHRED tires; whereas, in the absence of S9, the re-
verse was observed (see Table 7).
   The mutagenic potencies  of the particulate organics are
similar to those obtained from a variety of combustion emis-
sions,  although  the potencies of the  particulate organics
from the CHUNK tires in the presence  of S9 (approximately
10-12 rev//ig) are in the upper end of the range of values
typically found  for combustion emissions (13-17,33). The
greater mutagenic potencies in the presence  of S9 of the
particulate organics from CHUNK tires may have been due to
the greater initial burn rate that was achieved with CHUNK
tires. A greater burn rate might have produced more polycyclic
aromatic hydrocarbons (PAHs), which would be typical of the
class of compounds that require S9 in order to be mutagenic.
   The decreased mutagenic potency in the absence of S9 of the
particulate organics from CHUNK tires  might also have been
due to the greater burn rate  that was achieved with CHUNK
tires. A greate- burn rate might have resulted in the presence
of fewer substituted PAHs in the particulate organics from the

-------
4826
         TIRES, OPEN BURNING
Table 7. Mutagenic Potencies and Mutagenic Emission Factors for Tire Fire in TA98
Exp.
1
2
Condition
CHUNK
SHRED
CHUNK
SHRED
Bum
Rate,
kgfti
9.4
1.1
3.5
1.3
Fill.
EOM,
4.5
22.6
6.8
16.7
XAD (revertants per)
S9 fig
+ 2.2
1.9
+ 0.7
1.0
+ 2.1
1.4
+ 4.3
8.9
in3
1,536
1,326
623
890
1,573
1,049
4,726
9,781
kg of fuel X 10s
14.88
12.85
0.17
0.25
18.76
12.51
59.26
122.65
MJof
heat
42,690
36,866
487
717
53,822
35,891
170,016
351,822
«
12.0
2.3
4.3
8.9
10.4
2.3
7.0
7.0
Filter (revertants per)
m»
403,608
77,358
29,335
60,716
192,379
42,545
61,397
61,397
kgoffuelxlO'
873.61
167.44
484.23
100.25
1,036.52
229.23
769.25
769.25
MJof heat
2,506,387
480,385
1,389,255
2,875,455
2,973,775
657,660
2,206,978
2,206,978
 CHUNK tires. Thus, substituted PAHs would be more preva-
 lent in the participate organics from the SHRED tires, and sub-
 stituted PAHs are frequently mutagenic in the absence of S9;
 whereas PAHs require S9 in order to be mutagenic in the Sal-
 monella assay.

   Laboratory  Data:  Mutagenic Emission Factors.  Mutagenic
 emission factors are a convenient way to characterize muta-
 genic potency in terms of fuel consumed or heat produced by a
 combustion process. Mutagenic emiss. in factors of the DCM-
 extractable organics of tire burn effluent can be expressed as
 the mutagenic activity per unit weight of tires (revertants per
 kilogram of fuel) or per unit heating value of tires during com-
 bustion (revertants per megajoule of heat). These expressions
 are a function of the mutagenic potency of the organics (re-
 vertants per microgram), the burn rate (kilograms per hour),
 the flow rate (cubic meters per hour), and the percentage of
 DCM-extractable organics collected on the XAD or the filters.
 Table 7 shows the mutagenic emission  factors expressed by
 weight consumed and heat produced from effluent of burning
 CHUNK or SHRED tires.
   The  mutagenic   emission  factors  calculated  for  the
 semivolatile organics (XAD) from  CHUNK tires  in  the
 presence of S9 were greater than for SHRED tires in Day 1;
 however, this effect was reversed in Day 2 (see Table 7). This
 discrepancy is likely due to the different burn rates that oc-
 curred during the two experiments, as well as to other factors,
 such as material packing differences in the two experiments.
 For the semivolatiles, the mutagenic emission factors  were
 •comparable and, with the exception of the SHRED experiment,
 -'i.ey did not seem different in the presence or absence of S9.
 This could indicate that the mutagenicity of the semivolatiles
 was due equally  to PAH (S9-dependent) mutagenic activity
 and substituted PAH (S9-independent) mutagenic activity.
   The mutagenic emission factors for the particulate (filter)
 organics  from  CHUNK tires were greater than those  from
 SHRED tires in the presence of S9. In the  absence of S9,
 SHRED  tires  produced  higher mutagenic emission factors
 than did  CHUNK tires. This  suggests that the open burning
 of CHUNK tires  likely produced more PAH-type mutagens
 (S9 requiring) compared  to SHRED, and the open burning
 of SHRED tires likely produced more substituted  PAH-type
 mutagens (S9-independent)  than  did  CHUNK tires.  Two
 facts emerge from the data in Table 7: the DCM-extractable
particulate organics exhibit greater mutagenic potency (rever-
tants per microgram) compared to the semivolatile organics,
and the vast majority of the mutagenic activity (based on the
mutagenic emission factors) derives from the DCM-extractable
particulate organics, as opposed to the semivolatile organics.

   Laboratory Data: Role of Nitroarenes. Much of the mutagenic
activity of the particulate organics especially from the burns of
SHRED tires, did  not require S9 (see Table 7), which is typ-
ical of many nitroaromatics (31,32). The particulate organics
were evaluated for the presence of nitroaromatics by determin-
ing the mutagenic potency of the organics in strains TA98NR
and TA98/1,8-DNP{ of Salmonella. Because of limited amounts
of each sample, and the similar mutagenic potency profile ex-
hibited by the samples from the two experiments, a composite
sample was prepared composed of the particulate organics.
  Figure 6 shows the mutagenicity dose-response curves of
the composite particulate organics in the three strains. The
mutagenic potency of the organics was similar in strains TA98
(-S9) and TA98NR (-S9), 0.93 and 0.86 rev//tg, respectively.
Thus, little of the mutagenic activity of the organics was due to
the presence of mononitroaromatics. However, the mutagenic
potency of the  organics was reduced by 65% in strain TA98/
1,8-DNPg relative  to TA98, 0.36 vs 0.93 rev/^g, respectively.
This suggests that as  much as 65% of the mutagenic activity
as measured in the absence of S9 is due  to the presence of
dinitroaromatics or other nitroarenes or aromatic amines that
require metabolic conversion to arylhydroxylamines and then
esterification in order to be mutagenic.
                                           TA98 (+S9)
                        40       60
                        Dose, ug/plate
100
Figure 6. Mutagenicity dose-response curves of laboratory
composite sample of particulate organics in three strains of
Salmonella.

-------
                                                                                      TIRES, OPEN BURNING     4827
   Laboratory  Data:   Bioassay-Directed  Chemical   Analysis.
Chemical classes and individual chemicals that are respon-
sible for the mutagenic activity of the tire burn effluent
can  be determined  by characterizing the organics  using
bioassay-directed fractionation  and  chemical analysis (13).
The  resulting chromatograms and mutagrams of the four
particulate samples  were similar, and the average  mass
and  mutagenic activity recoveries  from the HPLC column
were 91 and 78%, respectively (19). Because of the  similarity
of the  mutagrams  and the limited amount of  each sample,
the unfractionated DCM extracts from the four particulate
samples were combined, and the composite sample was then
fractionated by HPLC as described previously.
   Figure 7 shows the mutagram of the composite sample of
the DCM-extractable particulate organics from the open burn-
ing of scrap rubber tires. The mutagram shows  four main ar-
eas of mutagenic activity, and more of the mutagenic activity
is direct-acting (-S9) than S9-dependent. The first  area of ac-
tivity coincides with the elution of four of the standards (naph-
thalene, benzo(a)pyrene, pyrene, and L-nitropyrene); the fifth
standard is acridine.
   Because some of the PAH standards eluted  in the  region
covered by fraction A (composed of fractions 2 and 3), fraction
A was analyzed for  the presence of various PAHs. Table 8
shows that 14 PAHs were identified in fraction A, confirming
the presence of this  class  of mutagen/carcinogen in the tire
effluent. PAHs require 89 in order to be mutagenic in Sal-
monella and, thus, could account for some of the  mutagenic
activity seen in the presence of 89 in fractions 2 and 3 of the
mutagram. However, there  is  also a  considerable amount
of S9-independent (—89)  mutagenic activity  in  these two
fractions that must  be due  to  other classes of compounds,
possibly requiring acetylation as suggested (see Fig. 6). Frac-
tion B. which was  composed of subfractions 22, 23, 24, and
25 from the mutagram, contained several oxygenated PAHs,
such as anthraquinone and  xanthone, which are  mutagenic
in strain TA98 (36,37). Fraction C, which was composed of
subfraction 43, contained phthalate. However, phthalates  are
ubiquitous in environmental samples and are not mutagenic
in Salmonella. Thus, another class or classes of compounds
are responsible for the S9-independent mutagenic activity in
subfraction C. Fraction D, which was composed of subfractions
47, 48, and 49, contained various polycyclic compounds with
ring nitrogens, such as acridine, which is also mutagenic in
Salmonella. Thus, a variety of aromatic, multiringed mutagens
were present in the particulate organics from the open burning
of tires.

   Ontario Field Samples: Mutagenic Potencies and Role of Ni-
troarenes. The mutagenic potency of the particulate organics
from the field sample was similar to that of the CHUNK tires
from the laboratory samples in TA98 in the absence of 89
(2 rev/jig) (19). Thus, the open burning of CHUNK tires in the
laboratory emitted particulate organics with a mutagenic po-
tency that most closely simulated that obtained from the open
burning of whole tires in the field.
   As with the laboratory composite sample, the field sample
showed little difference in mutagenic potency when evaluated
in strains TA98 and TA98NR (2.1 vs 2.4 rev/jig, respectively);
and the mutagenic potency of the field sample was reduced by
approximately 50% in strain TA98/l,8-DNPs relative to strain
TA98. Thus, as with the laboratory composite sample, much of
the mutagenic activity of the field sample particulate organics
as assayed in the absence of 89 was due to nitroarenes or aro-
matic amines that required conversion to arylhydroxylamines
followed by esterification in order to be mutagenic.
   Both the laboratory and field samples declined in muta-
genic potency during the months of storage at -80*C, exhibit-
ing a loss of approximately 50% activity during 1 year. Similar
results have been obtained with other combustion emissions,
such* as cigarette smoke condensate, which contain aromatic
amines but few mononitroaromatics. In contrast, highly stable
mutagenic activity is found with diesel exhaust, whose muta-
genic activity is due largely  to mononitroaromatics. Thus, the
presence of unstable, short-lived mutagenic species in the emit-
ted organics from the open burning of tires is consistent with
    1000
             5    10   15   20   25    30   35   40    45   50    55   60
                                  Fraction number            '
                      Figure 7. Mutagram of laboratory
                      composite sample of particulate
                      organics in strain TA98 using a
                      microsuspension Assay. Arrows
                      represent locations in which the
                      following standards eluted: 1,
                      naphthalene; 2, benzo (a)pyrene; 3,
                      pyrene; 4,1-nitropyrene; and 5,
                      Acridine, where - -S9 and - +S9.

-------
4828
          TIRES, OPEN BURNING
Table 8. Chemicals Identified in HPLC Fractions of Composite
of Tire Fire Samples                      	
Fraction0
Chemicals
A            Naphthalene, fluorene, phenanthrene, fluoranthene,
               pyrene, anthracene, benzo(a)anthracene, chrysene,
               benzo(6)fluoranthene, benzcWfluoranthene,
               benzo(a)pyrene, dibenzo(a,/i)anthracene,
               benzo(f,A.Operylene, indenot 1,2,3-ceOpyrene
B            Nonadecane, eicosane, anthraquinone, xanthor.e,
               benzan throne
C            Dioctyl phthalate
D            JJ-Caprolactam, cyclododecane, acridLne, naphthalic
               anhydride
             Benzanthrone, benzoisoquinoline, perinaphthenone,
               methylbenzocinnoline
• Fractions A-D were composed of the following fractions from the milligram
in Fig. 7: A (2.3), B (21-25). C (43), and D (47-49).
     500

     400

     300

     200

     100
                                        Q
                        „„  r-.n  r-,PH  ™-   r\
           Nl     N2    N3    N4    N5
                              Fraction
                                                                             N6     N7
                              Figure 8. Mutagenic potencies of neutral subfractions derived
                              from fraction A23 of the field participate organics assayed in
                              strain YG1024: D -  S9; 0 + S9.
the chemical composition of the samples inferred from their
mutagenic potency in strains TA98NR and TA98/1,8-DNP6.

   Ontario Field Samples: Bioassay-Oirected Chemical Analysis.
Because of the dependency of the field particulate organics
on the presence of O-acetyltransferase activity in order to be
mutagenic (as evidenced by the weak mutagenic potency of the
organics in strain TA98/1,8-DNP6, which lacks this enzyme),
the remaining studies of the field sample were performed in
strain YG1024, which has enhanced activity for  this enzyme
(34). Table 9 shows the mutagenic potencies of five fractions of
the field particulate organics in strain YG1024 in  the presence
and absence of 89. Most of the mutagenic activity is present in
fraction A23, which is a neutral fraction. After removal of any
residual aliphatics, this fraction was then subfractionated by
HPLC as described previously to produce seven  subfractions
whose mutagenic potencies are shown in Figure 8.
   Most of the mutagenic activity detected in the field sample
eluted in neutral subfraction  Nl, and most of  this activity
was S9-dependent, suggesting the presence of PAHs. The
mutagenic activity was greater  in the presence of S9 and
was dependent, somewhat on  0-acetylation. Substitution of
nitroreductase for O-acetyltransferase in the bacterial strains
greatly reduced the  mutagenic potency of the organics (data
not shown). GC/MS analysis of this subfraction confirmed the
presence of a variety of PAHs, many of which are mutagenic in
Salmonella (Table 10). Several of these PAHs were also present
in HPLC fraction A of the laboratory tire burn particulate
organics (see Table 8). A substantial amount of 89-dependent
mutagenic activity was present in another neutral subfraction
Table 9. Mutagenic Potency of Fractions of Field Tire Fire
Sample in YG1024         	
                              Table 10. Chemicals Identified in Fraction Nl of Ontario Field
                              Tire Fire Sample
                   Kevertants//ig
             Revertants/m3
Fraction
                 -S9
                             +89
                                         -S9
                        +S9
Al
A23
Aliphatics
A45
A6
0.0s
5.7
0.0"
1.8
0.0"
0.0-
17.8
0.0"
2.5
0.3
0.0"
122.0
0.0-
0.0°
0.0«
0.0«
350.0
19.4
146.0
2.7
                              l,12-Dimethylbenz(a)anthracene
                              1,8-Dunethylphenanthrene
                              1-Methy lbenz(a Janthracene
                              1-Methylpyrene
                              1-Phenylnaphthalene
                              ll-Benzo(a)fluorine
                              2,2'.Binaphthyl
                              2,7-Dimethylpyrene
                              2,8-Dimethylbenzothiophene
                              2-Ethylphenanthrene
                              2-Methylanthracene
                              2-Methylpyrene
                              2-phenylnaphthalene
                              5-Methylbenza(a)anthracene
                              7-Benz(d,e)anthrene
                              8-Methylbenz(a)anthracene
                              9-Ethylphenanthrene
                              Acephenanthrylene
                              AnthraC 1,2-6 Hhiophene
                              Anthracene
                              Benz(a)anthracene
                              Benzofa )f luorene
                              Benzo(a)pyrene
                              Benzo(6)naphtho< l,2-d)thiophene
                              Benzo(6)naphtho(2,l-d)thiophene
                              Benzo(c)phenanthrene
                              BenzoWpyrene
                              Benzo(g,A,t)perylene
                              Benzort Jfluoranthene
                              Benzof A )f luorenone
                              m-Terphenyl
                              Methylfluorene
                              p-Terphenyl
                              Pentacene
                              Phenanthrene
                              Phenanthro(9,10-b)thiophene
                              Pyrene
1 Values are <0.01.
(N7); however, this subfraction was not subjected to chemical
analysis.
   The unfractionated, whole particulate organic extract of the
field sample was also fractionated by HPLC in a manner identi-
cal to that used to fractionate the laboratory composite sample.
However, due to the decline in mutagenic potency at the time of
fractionation, the overall mutagenic potency of the field sample
was only 35% of that of the laboratory sample. Nonetheless, a
majority of the S9-dependent mutagenic activity eluted in the
PAH-containing fractions (fractions 2-4), similar to the labora-
tory composite sample (see Fig. 7).

   Comparison Between Laboratory and  Ontario  Field Samples.
PAH Concentrations. Because the results showed that PAHs
contributed substantially  to  the mutagenic activity of the
particulate organics emitted from the open burning of tires,
we determined the  concentrations of  selected PAHs in the
unfractionated, whole particulate organic extracts from the
laboratory and  field samples. For nearly all of  10 PAHs,
the concentration of a particular PAH in the  Ontario Field
Samples was within an order of magnitude of its concentra-

-------
                                                                                      TIRES, OPEN BURNING
                                                    4829
tion in the laboratory samples (Table 11). Because the burn
rate was known for the laboratory studies, the concentrations
of selected PAHs emitted per kilogram of tire was estimated
(see Table 11). With few exceptions, the condition of the tire
material (CHUNK vs  SHRED) had little  influence on the
concentrations of selected PAHs emitted in the  particulate
organics.
   The  two  samples  are  quite  similar;  however,  some
compounds  (eg,  pyrene  and  phenanthrene)  are present
at  higher relative  concentrations in  the Ontario  field
samples  than in  the  laboratory samples. This difference
might be due  to atmospheric transformation  reactions,
because  the  laboratory experiments  were conducted (and
the  sample collected) in a small building in the absence of
sunlight.
   Mutagenic Emission Factors. Because the Ontario field sam-
ples did not provide data on the burning rate of the tires, mu-
tagenic emission factors could not be calculated for the field
sample. Most of the mutagenic emission factors that have been
determined for other combustion emissions have been for the
DCM-extractable particulate organics evaluated in Salmonella
strain TA98 in the presence of 89. Because the mutagenic emis-
sion factors for the particulate organics in the presence of S9
were similar for CHUNK and SHRED tires, we calculated the
arithmetic average of the four values (two from each experi-
ment) to arrive at a single mutagenic  emission value in  the
presence of 89 for the open burning of tires. Figure 9 compares
this value to those of other  combustion emissions evaluated
in strain TA98 in the presence of 89. The various  combustion
emissions rank the same regardless of which way the muta-
genic emission factors are expressed.
   The mutagenic emission factor for open tire burning  is the
greatest of any other combustion emission studied previously
(see Fig. 9). For example, it is 3-4 orders of magnitude greater
than the  mutagenic emission factors  for the combustion of
oil, coal, or wood in utility boilers. The data labeled "kiln" in
Figure 9  are from small-scale  experiments  that examined
emissions from the combustion of the  listed compounds in a
laboratory-scale  rotary kiln incinerator  simulator that was
operated  under suboptimal conditions (16). The data labeled
"Dinoseb" are from pilot-scale studies that examined the effect
of nitrogen oxide (NO^-reducing combustion modifications on
the emissions from a boiler burning a  highly nitrated waste
(14). The  data labeled "Open Ag. Plastic Burning" are from
a controlled experiment examining emissions from the open
burning of agricultural plastic (15). All other data are from
field samples (13-17).
   The mutagenic emission factors for the residential combus-
tion of wood and for the open burning of agricultural plastic are
closest to that of the open burning of tires (see Fig. 9). Both of
these combustion conditions represent a type of open burning
similar to that of the tire burns in the present study. All three
conditions have poor combustion characteristics, and this is re-
flected in their high mutagenic emission factors. Thus, poor
combustion conditions result in elevated levels of PICs, which
result in elevated mutagenic  emission factors. It is clear from
the comparison in Figure 9 that the open burning of tires (as
well as of wood or plastic) results in exceptionally high muta-
genic emission factors. Thus, open burning, regardless of the
feed stock or fuel, results in greater mutagenic emission factors
than does controlled combustion provided by various types of
incinerators or boilers.
SUMMARY

This article reports on research efforts to characterize poten-
tially harmful emissions from the simulated open burning
of scrap tires, using laboratory and field data. Results were
analyzed from a chemical and a biological perspective.
  The laboratory simulation was necessarily crude, because it
would be extremely difficult to match the burning of the equiv-
alent of two tires with a 6 million tire, full-scale, stockpile fire.
Nevertheless, the study allowed the investigators to identify
and measure gaseous emissions and directly relate this infor-
mation to a mass burn rate. This task was accomplished by
accurately measuring dilution volumes, sample volumes, and
weights of tire material combusted.
  The dilution air added to the burn hut was used not only
to control known volumes introduced, but'also to  simulate
ambient conditions. It is  probable that the same  types of
compounds identified during this study are emitted during an
actual fire, but whether the average gaseous concentrations
and estimated emissions are comparable is uncertain. Com-
parison of PAH concentrations from laboratory samples agrees
quite well with PAH concentrations taken from a full-scale
tire fire in Ontario, Canada. Many of the same compounds
were  identified in actual plume  samples. Particularly good
agreement  exists in PAH plume measurements from the
Ontario fire. Reported plume concentrations of total PAHs
from  the field test are generally within the same  order of
magnitude as average gaseous concentrations observed during
the Laboratory tests. Measurements of CO and  metals also
indicate similar agreement. The lead and zinc measurements
show similar values in both absolute and relative concentra-
tions between the two metals. It may be reasonable to assume
thai the estimates'obtained during this simulation study are
within an order of magnitude of emissions realized from actual
stockpile fires.
  The results from the airborne metals portion of the study
were  inconclusive. Maximum values were presented, often
based oh detection levels.  Emissions  of lead and zinc may
reach significant quantities. Chemical analysis of ash residues
reveals that zinc composes nearly 50% of the total residue (14).
Evidently, the other metals known to be  contained in tires
remain in the ash residue. Although no attempt was made to
analyze ash residue, significant quantities of metals present
in the ash could potentially be leached out into groundwater
systems, posing another major problem.
  The values obtained by  the continuous emission monitors
for normal  combustion gases showed  that, as burn rate in-
creased, the amount of CO, S02, and unburned hydrocarbons
also  increased.  High  burn rate conditions were not fully
evaluated, so greater quantities of these gases,  particularly
SOo, may be emitted during a stockpile fire. Tires contain a
moderate amount of sulfur, so high emissions of SO2, while
likely only a minor contributor to the acid rain problem, could
have significant local consequences.
  The laboratory study was designed to identify the poten-
tial chemical hazards from tire fires on a small-scale, simula-
tion basis. It revealed the potential for the emission of large
amounts of organic compounds, primarily aromatics, some of
which are known to be hazardous. Although the estimates of
average gaseous concentrations and estimated emissions are
crude, the trends presented in regard to burn rate may be help-
ful in directing further research and control efforts. The fact

-------
4830
          TIRES, OPEN BURNING
                    ron^ntratlon of Selected PAHs in Tire Fire Emissions
p/g of Organic Extract*
Laboratory
PAH
Acenaphthalene
Acenaphthene
Anthracene
Benz(a)anthracene
Benzo(a)pyrene
Benzo(a./i)anthracene
Benzo(6)fluoranthene
Benzo(g,A,i)perylene
Benzo(%)fluoranthene
Chrysene
Fluoranthene
Fluorene
Indeno( l,2,3-crf)pyrene
Naphthalene
Phenanthrene
Pyrene
Field
C
t
337
1,389
833
179
853
397
304
2.341
2,797
e
179
C
3,333
3,631
CHUNK
3,678
23,181
1,825
7,764
8,554
81
6,721
5,843
7,117
7,414
25,052
1,324
5,210
81
5,256
2,676
SHRED
0
17,808
766
7.177
7.511
0
6,688
12,089
7,511
6,802
21,376
318
6,468
0
2,314
10,663
mgflcg of Tire*
Laboratory
CHUNK
861
290
56
82
85
1
70
66
74
71
339
261
52
816
238
34
.SHRED
562
2,446
50
102
114
0
88
159
99
92
458
187
86
486
253
152
           • Filter extracts only; XAD-2 extract data not reported; laboratory values are average of 2 experiments.
           * Data include both the filter and XAD-2 fractions.
           e Sample not analyzed for the presence of these compounds.
      S531
      3 •
  to
  a.
 O
I               Natural gas
     Cardboard and sorbent
              Polyethylene
                  Toluene
          Polyethylene/PVC
              Toluene/CCI4
                 Utility oil
                Utility coal
               Utility wood
             Stage dinoseb
   Stage and reburn dinoseb
             Residential oil
Municipal waste combustion
          Residential wood
    Open ag plastic burning
          Open tire burning
                            i 11 mill i 11 n ml  i 11 mill  i iimiil  i iiiiml  i inni
                                                                      tllllllll  It mini ii mini  iiiinnl  iiiinnl
io-3  io-2   lo-1   10°   iol    io2  io3  10°   lo1
             Rev/kg of fuel x IO5
                                                                                                        ininl
                                                                                   IO2
                                                                                      IO3   IO4
                                                                                       Rev/MJ
IO5    IO6   IO7
                   Figure 9. Mutagenic emission factors of combustion emissions in strain TA98 (+89).
                   Data for emissions other than tire burning are from Refs. 13-17.
that the SHRED condition resulted in a lower burn rate indi-
cates that the gaps between the tire material provide the ma-
jor avenue of oxygen transport. Oxygen transport appears to
be a major if not the controlling mechanism for sustaining the
combustion process. This fact could have advantageous impli-
cations for those attempting to combat tire fires. It may be
possible to fill the gaps between tires with a foam inhibitor, po-
tentially suffocating the fire from within.
   The same samples acquired in the laboratory experiments
were then analyzed using Salmonella-based bioassays. In gen-
eral, the semivolatile organics emitted from the open burning
                                                        of CHUNK or SHRED tires exhibited similar mutagenic poten-
                                                        cies. However, the mutagenic potencies of the particulate or-
                                                        ganic were greater than that of the semivolatiles regardless of
                                                        the size (CHUNK or SHRED). The open burning of CHUNK
                                                        tires produced more potent organics as assayed in the presence
                                                        of S9, but SHRED tiros produced more potent organics as as-
                                                        sayed in the absence of S9. This difference was likely a reflec-
                                                        tion of the different amounts or classes of chemicals produced
                                                        under the two combustion conditions.
                                                           Mononitroaromatics do not appear  to account for much of
                                                        the mutagenic activity of the particulate organics emitted from

-------
                                                                                            TIRES, OPEN BURNING
                                                        4831
 either the field or laboratory tire burns. However, other types
 of nitroarenes or aromatic amines appear to account for much
 of the mutagenic activity of these organics when assayed in the
 absence of S9. Most importantly, PAHs are present in similar
 amounts in both field and laboratory samples and appear to
 account for much of the mutagenic activity of both samples
 when assayed in the presence of S9. Based on the similar
 chemical composition and mutagenic  effects of the field and
 laboratory samples, it appears that the laboratory tire burn
 simulated reasonably well  a  major  real-world  tire  burn.
 This should  provide some confidence in  extrapolating  the
 results  from the  laboratory tire  burn  to real-world  tire
 burns.
   However, the neutral organic fractions from the laboratory-
 and Ontario field sample-derived burns did respond differently
 to enzyme-mediated oxidation. The nonpolar fractions from the
 laboratory sample (represented by the initial fractions in the
 mutagram in Fig. 7) were more mutagenic in the absence of 89
 than in the presence of S9. A roughly comparable fraction from
 the field burn (represented by subfraction Nl in Fig. 8) was
 more mutagenic  in the presence of 89. This difference in the
 mutagenic activity of the neutral organics from the two burns
 could be a result of the conversion of PAHs to nitrated-PAHs
 by atmospheric transformation, which would be more likely to
 have occurred in the field burn. For example, some nitrated-
 PAHs, such as 3-nitroperylene,  require metabolic  activation
 by S9 and O-acetyltransferase, but not by nitroreductase, in
 order to be mutagenic (38). The authors also showed that the
 metabolism of large,  nitrated-PAHs  by  mixed-function oxi-
 dases from rat liver yielded ring-oxidized compounds that were
 more readily  transported  into the bacteria than the  parent
 compounds.
   The vast majority of the mutagenic activity, as represented
 by the mutagenic emission factors, derives from the particu-
 late (condensed-phase) organics as opposed to the semivolatile
 organics. The mutagenic emission factor for the open burning
 of tires is the highest of any other combustion emission stud-
 ied previously. Although it is 3-4 orders of magnitude greater
 than that for the  combustion of oil, coal, or wood in utility boil-
 ers,  it is most similar to values for the open burning of wood
 and plastic. Open burning, regardless of feed stock or fuel, ap-
 pears to result in greater mutagenic emission factors than does
 controlled combustion as provided by various types of incinera-
 tors or boilers.
   On the basis of the high mutagenic potency and mutagenic
 emission factor associated  with the particulate organics emit-
 ted by open tire burning, the presence of many mutagens/car-
 cinogens (especially PAHs) in the particulate organics, and the
 high mutagenic emission factors for open burning in  general
 versus controlled combustion, we conclude that the open burn-
 ing of scrap rubber tires poses a potential environmental and
 health hazard. Because of the frequent occurrence of unwanted
combustion at tire disposal sites, and the potential environ-
 mental and health risks posed by such combustion, prudence
would suggest that such sites be reduced or eliminated in size
 and  number. Used tires may  be recycled, used in asphalt for
 roads, or incinerated under controlled conditions in combustion
devices for cogeneration  of power. Any of these uses would ap-
pear to be preferable to the environmental  and health risks
posed by the landfill disposal of tires  and the potential envi-
ronmental and health effects resulting from the open burning
of such tires.
ACKNOWLEDGMENTS
The initial tire burning study was performed by R. Rinehart and
K. Krebs of Acurex Corp. under EPA Contract No. 68-02-4701, Task
88-41. The authors would also like  to thank R. W. Williams and
L. R. Brooks (EPA) for their contributions to the HPLC fractionation
studies,  and D. Corr and N. Bonucore of McMaster University for
their work on the Ontario samples.
BIBLIOGRAPHY

 1. T.A. Sladek, Workshop on Disposal Techniques with Energy Re-
    covery for Scrapped Vehicle Tires, City and County of Denver,
    The Energy Task Force of the Urban Consortium for Technol-
    ogy Initiatives, US Department of Energy, Denver,  Colo., Feb.
    1987.
 2. US EPA Office of Solid Waste, Summary of Markets for Scrap
    Tires. EPA/530/SW-90-074b, Oct. 1991.
 3. A.T. Kearney Scrap Tire Use/Disposal Study, Prepared for the
    Scrap Tire Management Council, Sept. 1990.
 4. C. Clark, K. Meardon, and D. Russell Burning Tires for Fuel and
    Tire Pyrolysis: Air Implications, EPA-450/3-91-024 (NTIS PB92-
    145358). Dec. 1991.
 5. M. Pirnie, Air Emissions Associated with the Combustion of Scrap
    Tires for Energy Recovery, Prepared for the Ohio Air Quality De-
    velopment Authority, May 1991.
 6. Section 1038, H.R. 2950 Intermodal Surface Transportation Effi-
    ciency Act of 1991, First Session of the 102nd Congress, Enacted
    Dec. 18,1991.
 7. T.P. Livdahl, and M.S. Willey, Science 253,189-191 (1991).
 8. L.E. Haverfield,  and  B.L. Hoffman, Mosq. News 26, 433-435
    (1996).
 9. International Agency for Research on Cancer (IARC), Monographs
   •on the Evaluation of the Carcinogenic Risk of Chemicals to Hu-
    mans, The Rubber Industry, Vol. 28, IARC, Lyon, France, 1982.
10. T. Shibamoto, and C.-I.  Wei., Agric.  Biol. Chem,, 50, 513-514
    (1986).
11. J.V. Ryan,  Characterization of Emissions from the Simulated
    Open Burning of Scrap Tires,  EPA-600/2-89-054 (NTIS PB90-
    126004). Oct. 1989.
12. International Agency for Research on Cancer (IARC), Monograph
    Programme on the Evaluation of Carcinogenic Risk of Chemicals
    to Humans, Polynuclear Aromatic Compounds, Part 4, Bitumens,
    Coal-tar and Derivt   Products,  Shale Oils and Soots, Vol. 35,
    IARC, Lyon, France, 1985.
13. J. Lewtas, Fundam. Appl. Toxicol. 10,571-589 (1988).
14. D.M. DeMarini, V.S. Houk, J. Lewtas, R.W. Williams, M.G. Nish-
    ioka, R.K. Srivastava. J.V. Ryan, J.A. McSorley, R.E. Hall, and
    W.P. Linak, Environ. Sci. Technol. 25, 910-913 (1991).
15. W.P. Linak, J.V. Ryan, E. Perry. R.W. Williams, and D.M. De-
    Marini, J. Air Pollut. Control Assoc. 39,836-846 (1989).
16. D.M. DeMarini, R.W. Williams, E. Perry, P.M. Lemieux, and W.P.
    Linak, Combust. Sci. & Tech. 85,437-453 (1992).
17. R.R. Watts, P.M. Lemieux, RA.  Grote, R.W. Lowans, R.W.
    Williams, L.R.  Brooks, S.H. Warren,  D.M. DeMarini, D.A, Bell,
    and J. Lewtas, Environ. Health Persp. 98,227-234 (1992).
18. P.M. Lemieux, and J.V. Ryan J. AWMA 43,1106-1115 (1993)
19. P.M. Lemieux, and D.M. DeMarini,  Mutagenicity of Emissions
    from the Simulated Open Burning of Scrap Rubber Tires,  EPA-
    600/R-92-127 (NTIS PB92-217009). July 1992.
20. D. DeMarini, P. Lemieux. J. Ryan, L. Brooks, and R. Williams,
    Environ. Sci. and Tech. 28,136-141 (1994).
21. J. Kovach, Tire Business 7(22), 1, 32 (1990).

-------
4832

22. Analysis of Ambient Monitoring Data in the Vicinity of Open Tire
    fires, EPA-463/R-93-029 (NTIS PB94-156197), July 1993.
23. National Institute for Occupational Safety -nd Health (NIOSH),
    Health Hazard Evaluation Report, Rhinehart Tire Fire, Winches-
    ter, Virginia, HETA 84-049-1441, Mar. 1984.
24. Method  0030 in EPA SW-846 (NTIS PB88-239223) Test Meth-
    ods for Evaluating Solid Wastes, Vol. II, Field Manual Physical/
    Chemical Methods, 3rd ed., EPA, Nov. 1986.
25. Method 5040 in Ref. 24.
26. Method 0010, Appendix B in Ref. 24.
27. Federal Register Part VIII, 40 CFR Part 136, Vol 49, No. 209,
    Friday,  Oct. 26, 1984, Rules and  Regulations, Method  610-
    Polynuclear Aromatic Hydrocarbons, p. 43321.
28. B.L. Schulman, and  P.A. White, Reprint from the ACS Sympo-
    sium Series, #76, Solid Wastes and Residues, 1978.
29. Method 6010 in Ref. 24.
30. D.B. Maron, and B.N. Ames. Mutat. Res. 113,173-215 (1983).
31. E.C. McCoy,  M. Anders, and H.S.  Rosenkranz, Mutat. Res. 121,
    17-23 (1983).
32. M. Kohan, and L. Claxton, Mutat. Res. 124,119-200 (1983).
33. D.M. DeMarini, M.M. Dallas, and J. Lewtas, Teratogen. Carcono-
    gen. Mutagen. 9,287-295 (1989).
34. D.M. DeMarini, R.W. Williams, L.R. Brooks, and M.S. Taylor,
    Intern. J. Environ. Anal. Chem. 48,187-199 (1992).
35. M. Watanabe, M. Ishidate, Jr., and T. Nohmi, Mutat. Res. 234,
    337-348(1990).
36. E. Zeiger, B. Anderson, S. Haworth, T. Lawlor, and  K. Mortel-
    mans, Environ. Mol. Mutagen. 11 (Suppl. 12), 1-158 (1988).
37. M. Muramatsu, and T. Matsushima, Mutat. Res. 164, 274-275
    (1986).
38. J.G. Anderson,  D.R.  McCalla, D.W. Bryant, B.E. McCarry, and
    A.M. Bromke, Mutagenesis Z, 279-285 (1987).

-------