Emissions of polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans from the open burning of household waste in barrels

Article abstract of DOI:10.1021/es990465t

Backyard burning of household waste in barrels is a common waste disposal practice for which pollutant emissions have not been well characterized. This study measured the emissions of several pollutants, including polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans (PCDDs/PCDFs), from burning mixtures designed to simulate waste generated by a 'recycling' and a 'nonrecycling' family in a 208-L (55-gal) burn barrel at the EPA's Open Burning Test Facility. This paper focuses on the PCDD/PCDF emissions and discusses the factors influencing PCDD/PCDF formation for different test burns. Four test burns were made in which the amount of waste placed in the barrel varied from 6.4 to 13.6 kg and the amount actually burned varied from 46.6% to 68.1%. Emissions of total PCDDs/PCDFs ranged between 0.0046 and 0.48 mg/kg of waste burned. Emissions are also presented in terms of 2,3,7,8-TCDD toxic equivalents. Emissions of PCDDs/PCDFs appear to correlate with both copper and hydrochloric acid emissions. The results of this study indicate that backyard burning emits more PCDDs/PCDFs on a mass of refuse burned basis than various types of municipal waste combustors (MWCs). Comparison of burn barrel emissions to emissions from a hypothetical modern MWC equipped with high-efficiency flue gas cleaning technology indicates that about 2-40 households burning their trash daily in barrels can produce average PCDD/PCDF emissions comparable to a 182 000 kg/day (200 ton/day) MWC facility. This study provides important data on a potentially significant source of emissions of PCDDs/PCDFs. Backyard burning of household waste in barrels is a common waste disposal practice for which pollutant emissions have not been well characterized. This study measured the emissions of several pollutants, including polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans (PCDDs/PCDFs), from burning mixtures designed to simulate waste generated by a 'recycling' and a 'nonrecycling' family in a 208-L (55-gal) burn barrel at the EPA's Open Burning Test Facility. This paper focuses on the PCDD/PCDF emissions and discusses the factors influencing PCDD/PCDF formation for different test burns. Four test burns were made in which the amount of waste placed in the barrel varied from 6.4 to 13.6 kg and the amount actually burned varied from 46.6% to 68.1%. Emissions of total PCDDs/PCDFs ranged between 0.0046 and 0.48 mg/kg of waste burned. Emissions are also presented in terms of 2,3,7,8-TCDD toxic equivalents. Emissions of PCDDs/PCDFs appear to correlate with both copper and hydrochloric acid emissions. The results of this study indicate that backyard burning emits more PCDDs/PCDFs on a mass of refuse burned basis than various types of municipal waste combustors (MWCs). Comparison of burn barrel emissions to emissions from a hypothetical modern MWC equipped with high-efficiency flue gas cleaning technology indicates that about 2-40 households burning their trash daily in barrels can produce average PCDD/PCDF emissions comparable to a 182 000 kg/day (200 ton/day) MWC facility. This study provides important data on a potentially significant source of emissions of PCDDs/PCDFs.

Full text of DOI:10.1021/es990465t

Environ. Sci. Technol. 2000, 34, 377-384
Introduction
Emissions of Polychlorinated
Dibenzo-p-dioxins and
Open Burning of Household Waste. In many areas of the
country, residential solid waste disposal practices consist of
open burning using barrels or other similar devices instead
of, or in addition to, disposal in municipal landfills or
municipal waste combustors (MWCs). The motivations for
households that open burn their garbage may include
convenience, habit, or waste disposal cost avoidance (1).
Modern refuse combustors have tall stacks, specially designed
combustion chambers, and high-efficiency flue gas cleaning
systems that serve to minimize the impact of emissions
associated with waste combustion. In contrast, emissions
from open burning of residential solid waste are released at
ground level, possibly resulting in decreased dilution by
dispersion. Additionally, the low combustion temperature
and locally oxygen-starved conditions associated with back-
yard burning may result in incomplete combustion and
increased pollutant emissions.
Polychlorinated Dibenzofurans from
the Open Burning of Household
Waste in Barrels
P A U L M . L E M I E U X *
National Risk Management Research Laboratory,
United States Environmental Protection Agency,
Research Triangle Park, North Carolina 27711
C H R I S T O P H E R C . L U T E S
ARCADIS Geraghty & Miller, Inc., 4915 Prospectus Drive,
Durham, North Carolina 27713
Only two previous studies characterized emissions as-
sociated with open burning of residential refuse in a backyard
burner [e.g., a 208-L (55-gal) drum], and neither was published
in the peer-reviewed literature (1, 2). Both study designs
included a hood and stack constructed above the 208-L (55-
gal) drum to capture the plume and facilitate pollutant
emissions tests.
J U D I T H A . A B B O T T
Bureau of Toxic Substance Assessment, New York State
Department of Health, Albany, New York 12203
K E N N E T H M . A L D O U S
Wadsworth Center for Laboratories and Research,
New York State Department of Health,
Albany, New York 12203
The Two Rivers Regional Council of Public Officials
(
Illinois) study estimated emissions per unit mass of waste
initially present in the barrel for several different air toxics
[
(
e.g., total PCDDs/ PCDFs, total volatile organic compounds
measured as methane), air toxics metals (antimony, arsenic,
Backyard burning of household waste in barrels is a
common waste disposal practice for which pollutant emissions
have not been well characterized. This study measured
the emissions of several pollutants, including polychlorinated
dibenzo-p-dioxins and polychlorinated dibenzofurans
barium, beryllium, cadmium, chromium, lead, manganese,
mercury, phosphorus, and titanium), particulate matter (PM),
hydrogen chloride (HCl), carbon monoxide (CO), and oxides
of sulfur and nitrogen] associated with open burning of
residential solid waste. This study included a survey of 187
residents in rural counties of Illinois to determine the quantity
and type of wastes burned, the management of the ash, and
the motivation for burning. The Western Lake Superior
Sanitary District (Minnesota) study evaluated emissions of
(PCDDs/PCDFs), from burning mixtures designed to simulate
waste generated by a “recycling” and a “nonrecycling”
family in a 208-L (55-gal) burn barrel at the EPA’s Open
Burning Test Facility. This paper focuses on the PCDD/
PCDF emissions and discusses the factors influencing PCDD/
PCDF formation for different test burns. Four test burns
were made in which the amount of waste placed in the barrel
varied from 6.4 to 13.6 kg and the amount actually
burned varied from 46.6% to 68.1%. Emissions of total PCDDs/
PCDFs ranged between 0.0046 and 0.48 mg/kg of waste
burned. Emissions are also presented in terms of 2,3,7,8-
TCDD toxic equivalents. Emissions of PCDDs/PCDFs appear
to correlate with both copper and hydrochloric acid
emissions. The results of this study indicate that backyard
burning emits more PCDDs/PCDFs on a mass of refuse
burned basis than various types of municipal waste
combustors (MWCs). Comparison of burn barrel emissions
to emissions from a hypothetical modern MWC equipped
with high-efficiency flue gas cleaning technology indicates
that about 2-40 households burning their trash daily in
barrels can produce average PCDD/PCDF emissions
comparable to a 182 000 kg/day (200 ton/day) MWC facility.
This study provides important data on a potentially
significant source of emissions of PCDDs/PCDFs.
2
,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD); no other pol-
lutants were measured. The results of these two studies
indicated that the estimated emissions per unit mass of waste
initially present in the barrel were higher than for controlled
incinerators, with the Illinois study suggesting a factor of 17
higher than incinerators for total PCDDs/ PCDFs and the
Minnesota study suggesting a factor of 20 higher than
incinerators for 2,3,7,8-TCDD.
In another study, emissions were quantified from the
burning of municipal refuse on a burn table equipped with
a cone to funnel the pollutants to a sampling port (3). Samples
were analyzed for several combustion gases and hydrocar-
bons, including polycyclic aromatic hydrocarbons (PAHs),
but not PCDDs/ PCDFs. This study did not simulate the
oxygen-starved conditions typically found in a burn barrel.
While these studies provide useful information, there are
limitations associated with their results (e.g., the uncertainty
in the dilution air and a limited number of target compounds).
This paper presents a portion of the results of a study to
qualitatively identify and quantitatively measure the emis-
sions from open burning of residential solid waste in burn
barrels using techniques that would minimize the limitations
of previous studies. The full results of this study are contained
in a U.S. Environmental Protection Agency (EPA) report (4).
Form ation of PCDDs/PCDFs. Extensive research has been
*
Corresponding author phone: (919)541-0962; fax: (919)541-0554;
e-mail: lemieux.paul@epa.gov.
conducted to investigate the formation mechanisms of
1
0.1021/es990465t CCC: $19.00

2000 Am erican Chem ical Society
VOL. 34, NO. 3, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3 7 7
Published on Web 01/04/2000

PCDDs/ PCDFs in the years since PCDDs/ PCDFs were first
discovered in the exhaust gases from MWCs in 1977 (5). Since
then, multiple formation mechanisms have been proposed
TABLE 1. Composition of Waste
avid
(
6-12).
Field studies on MWCs have shown that the amount of
nonrecycler recycler
(%)
(%)
fly ash (and its accompanying metallic catalysts) and organic
precursors that pass through the temperature window
between 250 and 700 °C as well as the amount of time spent
in that optimal temperature window are the primary variables
affecting PCDD/ PCDF emissions (13). However, field studies
have been unable to clearly demonstrate a correlation
between chlorine input and PCDD/ PCDF emissions in full-
scale combustion systems, probably because other variables
paper
newspaper, books, and office paper
m agazines and junk m ail
corrugated cardboard and kraft paper
paperboard, m ilk cartons, and drink
boxes
32.7
11.1
7.6
3.3
10.3
61.9
plastica
PET no. 1
0.6
6.6
0.2
0.1
0.1
5.7
3.7
1.1
HDPE no. 2, LDPE no. 4, and PP no. 5
PVC no. 3
PS no. 6
m ixed no. 7
food waste
textile/leather
wood (treated/untreated)
glass/ceram ics
bottles/jars
ceram ics (broken plates and cups)
m etal, ferrous
iron, cans
non-ferrous
alum inum (cans, foil)
other non-iron (wire, copper pipe,
batteries)
10.4
4.5
0.3
(
PM carryover, PM control device temperature, and com-
bustion efficiency) dominate (14). It is the authors’ view that
chlorine is present in excess relative to the other reactants
in these systems.
On the basis of current PCDD/ PCDF formation theories,
a “worst-case scenario” for the formation of PCDDs/ PCDFs
from combustion systems would combine the following
features:
0.3
3.7
9.7
0.4
6.9
4.0
(
(
(
(
(
(
(
i) poor gas-phase mixing
ii) low combustion temperatures
iii) oxygen-starved conditions
iv) high PM loading
v) PM-bound copper
vi) presence of HCl and/ or chlorine
vii) significant gas-phase residence time in the 250-700
7.3
1.7
1.1
1.0
3.7
% total
100
100
°
C temperature range
a
PET, poly(ethylene terephthalate); HDPE, high-density polyethylene;
LDPE, low-density polyethylene; PP, polypropylene; PVC, poly(vinyl
chloride); and PS, polystyrene.
Open burning is a unique combustion source that appears
to fit all of the requirements for such a worst-case scenario
for the production of PCDDs/ PCDFs. A burn barrel provides
poor combustion conditions (providing a source of organic
precursors). This paper will show that a burn barrel also
provides significant gas-phase residence time in the optimal
PCDD/ PCDF formation temperature window in the presence
of significant amounts of HCl and a plentiful supply of PM.
Focus of Paper. The main focus of this paper will be to
examine a subset of the results from the EPA burn barrel
tests: the emissions of PCDDs/ PCDFs from open burning of
household waste in barrels. These data will be presented in
an effort to explain the magnitudes of the PCDD/ PCDF
emissions as well as the reasons for the differences between
the PCDD/ PCDF emissions measured under the different
test conditions. Additionally, a comparison of burn barrel
emissions to emissions from various types of MWCs will be
presented.
as “miscellaneous” were included by adding that percentage
to the item(s) within each category that possessed the largest
percentage(s). For the plastic category, the miscellaneous
percent was relatively large and was divided between the
two largest percentages within that category [for the avid
recycler it was divided between poly(vinyl chloride) (PVC)
and LDPE and for the nonrecycler it was divided between
HDPE and LDPE]. Although this simplification was dealt with
in a consistent manner, the percentage of PVC for the avid
recycler rose from 0.7% to 4.5%, which was greater than
expected. Table 1 presents the composition of the simulated
waste streams for the two scenarios. The simulated household
waste was prepared primarily from raw materials diverted
from the household wastes of ARCADIS staff members.
Moisture content of the waste was not measured or altered.
Household hazardous wastes (e.g., household chemicals,
paint, grease, oils, tires, and other vehicle parts) were not
included in the waste burned.
Experim ental Conditions. Initial charge mass for each
test burn was determined by loading the barrel to the same
level for each test. Thus, approximately the same volume
(uncompacted) of waste was used for each test burn. Since
the nonrecycler’s waste was less dense than the avid recycler’s
waste, the test burns for the nonrecycler had a smaller initial
waste mass. There were two test burns for each scenario,
and these were performed in EPA’s Open Burning Test Facility
(4). The mass burned ranged from 6.4 to 13.6 kg (14-30 lb)
per test. The burn device was a 208-L (55-gal) steel drum,
modified for ventilation by placing 24 1.27-cm (0.5-in.)
diameter holes evenly spaced around the bottom of the barrel
(as informal observations suggest is typically done by
homeowners). The barrel was sandblasted prior to use to
remove paint in an effort to simulate the use of a weathered,
used burn barrel that would represent the most common
residential burn device as well as to remove any potential
contaminants that would bias emission measurements. The
barrel was placed on an electronic scale platform to allow
the mass consumed by combustion to be monitored. A
Experimental Section
General. The objective of this project was to qualitatively
identify and quantitatively measure the emissions of numer-
ous pollutants from the open burning of household waste in
barrels. Analytical work was divided between ARCADIS
Geraghty & Miller (ARCADIS) and the New York State (NYS)
Department of Health, Wadsworth Center for Laboratories
and Research (WCL&R).
Due to the highly variable nature of household waste
generation, a reasonable representation of a waste stream
for disposal in a burn barrel was prepared based on the typical
percentages of various materials characterized and quantified
for NYS residents. The characterization was performed by
the NYS Department of Environmental Conservation, Divi-
sion of Solid Waste. On the basis of this detailed charac-
terization, waste-stream scenarios for an “avid recycling”
and a “nonrecycling” family of four were simulated for this
project. For each scenario, the waste composition was
simplified by combining the percentages for similar materials
[e.g., the percentages for high-density polyethylene (HDPE),
low-density polyethylene (LDPE), and polypropylene (PP)
were combined]. For each waste category, items characterized
3
7 8
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 3, 2000

TABLE 2. Run Duration and Burn Mass
start final mass
amt
mass mass burned burned duration
test no.
test
(kg) (kg)
(kg)
(%)
(min)
1
2
3
4
5
avid recycler 12.5 4.4
avid recycler 13.6 4.4
8.1
9.2
0.0
3.3
4.1
65.3
68.1
77
83
92
62
91
hut blank
0.0 0.0
6.4 3.1
8.8 4.7
nonrecycler
nonrecycler
51.6
46.6
the expected high concentrations of analytes in these tests,
this sampler was operated at approximately 28.3 L/ min (1
ft / min) for approximately 1.5 h. The temperature of air
3
FIGURE 1. Burn hut diagram (numbers represent thermocouple
locations).
entering the train and within the PUF cartridge was assessed
during preliminary tests in order to decide if further
precautions were necessary to cool the system. Since further
precautions were required, a copper cooling coil was
fabricated to enclose the exterior of the PUF module. The
method of operation of this sampling train differs from
method TO-9 in other respects:
(i) Due to constraints of facility size, the sampler location
criteria in TO-9 were modified (i.e., the sampler was located
inside the burn hut).
schematic diagram of the burn hut portion of the Open
Burning Test Facility is shown in Figure 1. The burn hut is
a small outbuilding 335 cm long (132 in.), 272 cm wide (107
in.), with a peak height of 221 cm (87 in.). High-volume air
handlers provided metered dilution air into the burn hut
yielding approximately 2.46 air exchanges per minute.
Additional fans were set up inside the burn hut to enhance
mixing within the hut. Samples for the continuous emission
monitors (CEMs) were collected from the sampling duct.
Other sampling devices were set up and operated like ambient
air samplers inside the burn hut.
(ii) The flow through the sampler was measured by a
separate dry gas meter rather than a Venturi and Magnehelic
gauge as discussed in TO-9.
Before each test, the material to be combusted was placed
in the barrel; air flow through the facility was initiated;
temperature measurements with thermocouples were begun;
and 15 min of background data was obtained on CEMs for
(iii) Analysis was performed using high-resolution gas
chromatography and low-resolution mass spectrometry
(HRGC/ LRMS) using EPA Methods 23 and 8280 (19, 20).
(iv) The filter and vapor-phase module were extracted
and analyzed together.
2 2
oxygen (O ), carbon dioxide (CO ), CO, and total hydrocar-
bons (THCs). The material to be combusted was then ignited
for a short period (<3 min) using a propane torch. Sampling
was initiated at least 2 min after the removal of the propane
These samples were spiked, extracted, and concentrated
by ARCADIS. The extracts were shipped on ice to WCL&R for
analysis by EPA Method 8280.
flame. Propane flames generally produce only water, CO
2
,
Estim ated Em issions. Estimated emissions of PCDDs/
PCDFs per unit mass consumed by combustion were
calculated by assuming thorough mixing of air inside the
burn hut and using
and small quantities of low molecular weight products of
incomplete combustion. Since these products were expected
to have largely dissipated prior to initiation of sampling, this
procedure should not have biased the results. This assump-
tion was verified by a hut blank experiment during which the
propane torches were lit, but household waste was not
burned. Additionally, the hut blank experiment provided
information for the assessment of background contaminant
concentrations in the ambient air that was pumped through
the facility. A single integrated sample was collected over the
course of each active burn, and sampling was terminated
when the burn mass did not change over an extended period.
Sam pling and Analysis. Metals were sampled in ac-
cordance with Method 101A (15) modified to be nonisokinetic
since sampling was not done from a duct. At WCL&R, the
metals were extracted from the particulates, and the extracts
were analyzed using inductively coupled plasma mass
spectrometry, atomic absorption (AA) spectrophotometry,
electrothermal AA spectrophotometry (graphite furnace), and
cold vapor AA using EPA standard methods.
HCl was sampled and analyzed in general accordance
with EPA Method 26 (16), except that isokinetic sampling
procedures were not utilized.
PM with an aerodynamic diameter <2.5 µm (PM2.5) was
measured using a dichotomous sampler placed inside the
burn hut (17).
PCDDs/ PCDFs were sampled using a Graseby PS-1
sampler operated within the burn hut. This train, designed
to comply with the EPA’s ambient sampling method TO-9
E ) (CsampleQhuttrun)/ (mburned
)
(1)
where Eis the estimated emissions in milligrams per kilogram
mg/ kg) of waste burned, Csample is the concentration of the
pollutant in the sample in milligrams per cubic meter of air
(
3
(
mg/ m ), Qhut is the flow rate of dilution air into the burn hut
3
in cubic meters per minute (m / min), trun is the run time in
minutes, and mburned is the mass of waste consumed by
combustion over the duration of the test burn in kilograms.
Results
CEM and Therm ocouple Data. Table 2 lists the burn masses
and duration for the four run conditions and the hut blank.
Table 3 lists the CEM and thermocouple data from the four
runs that burned waste. Thermocouple locations are labeled
in Figure 1. Average data are reported as the mean across the
entire run duration from ignition until sampling ceased. Note
2 2
that O and CO data are not reported in Table 3 because
they were very nearly at ambient concentrations over the
entire run duration. Figure 2 shows the time/ temperature
history of thermocouple 8, which was mounted near the top
opening of the barrel. Note that the temperatures are in the
250-700 °C temperature window for a significant portion of
the run time.
Statistical analyses were performed using the JMP software
to examine whether the average temperature over the run
or the maximum temperature over the run gave a correlation
with PCDD/ PCDF emissions. No statistically significant
correlations were found. It is likely that temperature has an
(
18), consists of an open-faced filter holder followed by a
polyurethane foam (PUF)-sandwiched XAD-2 bed vapor trap.
Because this sampler does not have a particulate size
separation device, fairly low flow rates can be used. Given
VOL. 34, NO. 3, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3 7 9

TABLE 3. CEM and Thermocouple Data Averaged over Run
Duration
run 1,
avid
run 2,
avid
run 4,
run 5,
recycler recycler
nonrecycler nonrecycler
THC (ppm )
CO (ppm )
8.1
10.0
6.5
7.3
114.2
56.6
na
110.8
161.7
54.7
53.4
192.5
44.5
10.4
14.1
78.6
51.8
74.7
na
317.2
59.9
48.4
107.7
38.1
11.1
15.4
74.4
50.0
71.0
na
592.2
52.7
50.0
109.7
40.5
tem p 1 (°C)
tem p 2 (°C)
tem p 3 (°C)
tem p 4 (°C)
tem p 5 (°C)
tem p 6 (°C)
tem p 7 (°C)
tem p 8 (°C)
tem p 9 (°C)
101.1
58.0
a
na
69.5
276.1
64.3
53.9
117.7
44.0
a
na, not available (therm ocouple failure).
FIGURE 3. Congener distribution of PCDDs/PCDFs. Results are
averaged across duplicate run conditions with nondetects set to
zero. Note that OCDD data are suspect due to the presence in hut
blank samples and are not shown.
between 40 and 120% being desirable. The calculations
correcting for standard recovery can lead to quantitation
errors when recoveries are outside this range. The complete
recovery data can be found in the full report (4).
Figure 3 illustrates the distribution of congener emissions
across all runs. PCDFs were higher than PCDDs, which is
consistent with results seen from MWCs, hazardous waste
incinerators (HWIs), and other combustion sources (14). In
general, PCDDs/ PCDFs emissions were higher for the avid
recycler than for the nonrecycler. Chlorobenzene emissions
also exhibited this same general trend (4), although a strong
correlation between chlorobenzenes and PCDDs/ PCDFs was
not seen, although chlorobenzenes are believed to be one of
the primary organic precursors leading to formation of
PCDDs/ PCDFs. The high PCDD/ PCDF emissions of run 1
strongly influenced the average value for the avid recycler.
FIGURE 2. Temperature vs time.
Discussion
Differences between Runs. Household waste is an extremely
heterogeneous material; its composition varies greatly among
households and also week by week for individual households.
These tests were performed on a simulated waste stream
intended to represent two types of households, with two
samples of each household’s waste. This prevented statistical
examination of relationships between waste composition,
combustion conditions, and pollutants that were emitted.
However, it is still possible to gain an understanding of
potentially important factors that contribute to the emissions
of PCDDs/ PCDFs from open burning of garbage.
influence on emissions of PCDDs/ PCDFs; however, given
the transient nature of the experiments and the limited
number of data points, no effect was discernible from these
data.
PCDD/PCDF Data. Table 4 lists the estimated emissions
for PCDDs/ PCDFs for each test burn. Where no peak was
detected above the detection limit, estimated emissions were
calculated based on the method detection limits and reported
as “less than” values. For the most part, emissions of PCDDs
(
but not PCDFs) were below detection limits for the non-
recycler, except for OCDD, which was higher than that of the
avid recycler. The hut blank sample showed levels of OCDD
comparable to the combustion runs, and the nonrecycler
runs also exhibited high levels of OCDD while other PCDDs/
PCDFs were low. Subsequent informal laboratory audits have
suggested the presence of native OCDD in the supposedly
isotopically labeled spiking mixture that was used at the
APPCD analytical laboratory during that time. For this reason,
OCDD data should be treated as suspect; therefore, all
discussions are based on the PCDD/ PCDF data with the
OCDD data excluded. It must be noted that some internal
standard recoveries were outside of the generally accepted
range between 40 and 120%, so quantitation for some
congeners may be questionable, although qualitatively the
data are sound. PCDD/ PCDF results are corrected for
recovery of internal standards, with recoveries ranging
There are several possible explanations for the observation
that the avid recycler’s emissions of PCDDs/ PCDFs were
higher than those of the nonrecycler. As pointed out above,
the procedure for preparing waste samples yielded a PVC
content of 4.5% for the avid recycler’s waste versus 0.2% for
the nonrecycler. The higher proportion of PVC plastic in the
avid recycler’s waste stream could potentially increase
formation of chlorinated organic compounds. Combustion
conditions such as time/ temperature history, mixing pat-
terns, and oxygen availability as well as the particular mixture
of carbon molecules and chlorine in the presence of a metal
catalyst are all important factors in the formation of PCDDs/
PCDFs. All of these variables potentially changed between
the avid recycler and the nonrecycler test cases.
Another possible explanation for the differences in
emissions between run conditions and between runs at the
3
8 0
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 3, 2000

TABLE 4. Estimated Emissions of PCDDs and PCDFs (mg/kg)
avid recycler
nonrecycler
isomer
compound
TCDD
run 1
run 2
run 4
run 5
2
1
1
1
1
1
1
2
1
2
1
1
2
1
1
1
1
378
2378
<0.0009
0.0013
0.0002
0.0014
0.0008
0.0153
0.0115
0.0022
0.0035
0.0013
0.0012
0.0067
0.0094
0.0024
0.0439
0.0004
0.0114
0.0141
0.0191
0.0099
0.0338
0.0115
0.158
<0.0005
<0.0005
<0.0001
<0.0006
<0.0006
0.0008
0.0005
0.0002
<0.0004
0.0002
0.0001
0.0006
0.0009
0.0003
0.0015
<0.0003
0.0005
0.0018
0.0008
0.0004
0.0015
0.0005
0.0224
0.0106
0.0056
0.0021
0.0005
0.005
<0.0003
<0.0003
<0.0004
<0.0006
<0.0005
<0.0006
0.0448
0.0001
0.0001
0.0002
0.0001
0.0002
0.0001
<0.0004
0.0002
<0.0005
<0.0007
<0.0003
<0.0003
<0.0006
<0.0006
0.0448
<0.0003
<0.0003
<0.0003
<0.0005
<0.0004
<0.0005
0.0317
<0.0003
<0.0002
<0.0003
0.0001
<0.0003
<0.0003
<0.0003
0.0034
<0.0003
<0.0006
<0.0003
<0.0003
<0.0005
<0.0005
0.0317
0.0007
<0.0002
0.0005
0.0034
<0.0006
0.0317
0.0046
0.0046
0.000903
PECDD
HXCDD
HXCDD
HXCDD
HPCDD
23478
23678
23789
234678
2346789
378
a
OCDD
TCDF
2378
3478
PECDF
PECDF
HXCDF
HXCDF
HXCDF
HXCDF
HPCDF
HPCDF
OCDF
23478
23678
34678
23789
234678
234789
2346789
total
total
total
total
total
total
total
total
total
total
total
total
total
total
TCDD
PECDD
HXCDD
HPCDD
a
OCDD
TCDF
0.0038
0.0024
0.0011
0.0002
<0.0007
0.0448
0.0075
0.0075
PECDF
HXCDF
HPCDF
OCDF
PCDD
PCDF
0.0995
0.0781
0.0576
0.0114
0.0884
0.4046
0.4815
0.0054
0.0412
0.0457
0.00123
b
PCDD/PCDF
c
TEQ
0.000759
a
OCDD results are suspect due to presence of elevated levels in the blanks. b Without OCDD results included. c TEQ, toxic equivalency (25).
TEQ contributions from non-detects were calculated at the detection lim it and OCDDs were neglected.
TABLE 5. Estimated Emissions of PCDDs/PCDFs, Copper, Hydrogen Chloride, Chlorobenzenes (CBs), Chlorophenols (CPs), and PM2.5
(mg/kg)
run
PCDDa
PCDF
total PCDD + PCDF
Cu
HCl
CBs
CPs
PM2.5
run 1, avid recycler
run 2, avid recycler
run 4, nonrecycler
run 5, nonrecycler
0.0769
0.0045
nd
0.4046
0.0412
0.0075
0.0046
0.4815
0.0457
0.0075
0.0046
15.01
6.18
2.16
3280
1510
481
0.287
1.73
0.416
0.432
0.576
1.79
0.413
1.5
6930
3580
20070
14800
b
nd
0.573
86.4
a
OCDD m easurem ents are not included in total of PCDDs. b nd, none detected.
same condition may be a cause due to the heterogeneous
nature of the household waste itself. Despite the efforts to
create an artificial waste stream that simulated “real”
household waste, duplicated for each run condition, there
undoubtedly were slight differences between the batches
used for the runs, such as bleaching agents, coatings, inks,
dyes, or foreign matter, or how the material was stacked in
the barrel. One possible explanation can be inferred from
examining the metals and HCl emissions.
chlorophenols were also compared to emissions of PCDDs/
PCDFs, and no correlations were found.
Note that run 1 exhibited higher copper and HCl emissions
than the other runs. Specifically, the emissions of copper,
HCl, and PCDD/ PCDF for run 1 were higher than those for
run 2 by factors of approximately 2, 2, and 10, respectively.
These observations provide a possible explanation for the
higher PCDD/ PCDF emissions observed during run 1 and
are consistent with current theories of heterogeneous dioxin
formation.
Effect of Exhaust Gas Constituents. On the basis of
current PCDD/ PCDF formation theory, it would be expected
that exhaust gas constituents such as PM, copper, or HCl
might affect PCDD/ PCDF emissions. Table 5 lists the total
PCDD +PCDF emissions (excluding the suspect OCDD data),
Figure 4 shows emissions of PCDDs/ PCDFs as a function
of copper and HCl emissions. These data suggest that
increases in emissions of either copper or HCl may be
associated with increases in PCDD/ PCDF emissions. The
chlorine- and copper-containing fractions of the waste may
have been distributed in the waste mixture in a manner that
introduced these elements into the combustion zone of the
burn barrel, thereby making copper and HCl available as
reactants for PCDD/ PCDF formation. Such observations are
consistent with currently accepted theories of heterogeneous
PCDD/ PCDF formation from combustion of municipal waste.
More conclusive statements about the relationship between
copper emissions, HCl emissions, and emissions of PM2.5
.
Although PM with an aerodynamic diameter less than 15 µm
was also measured, greater than 85% of the PM was less than
2
.5 µm in diameter. PM2.5 emissions did not correlate with
PCDD/ PCDF emissions. It may be that there was an excess
of PM available in all run conditions, with the availability of
solid-phase surface area not being a limiting factor. Other
measured metallic species as well as chlorobenzenes and
VOL. 34, NO. 3, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3 8 1

-
6
burn barrel study reported 4.0 × 10 mg of 2,3,7,8-TCDD/
kg of garbage fed into the barrel. The 2,3,7,8-TCDD emission
estimate presented in this paper was based on the method
-
4
detection limit and was estimated to be <5 × 10 mg/ kg of
trash consumed by combustion. The Illinois study reports
-
4
6
.2 × 10 mg of total PCDD/ PCDF/ kg of refuse fed into the
barrel (average of three test burns). These results can be
directly compared to results presented in this paper by
adjusting for the percent of mass reduced during each test
-
3
burn. The adjusted average emissions are 1.6 × 10 mg of
total PCDDs/ PCDFs/ kg of refuse consumed by combustion
(
average of three test burns). These Illinois data compare
-
4
well with the PCDD/ PCDF emissions factors of 7.6 × 10
-
-
3
4
.6 × 10 mg of total PCDD/ PCDF (excluding OCDD)/ kg of
refuse consumed by combustion presented in this paper for
the nonrecycler scenario.
Com parison with Em issions from Incineration. To place
burn barrel emissions in perspective, estimated emissions
for PCDDs/ PCDFs from burn barrels were compared to
average emission factors for three types of incinerators [mass
burn, modular, and refuse-derived fuel (RDF) combustors]
FIGURE 4. PCDD + PCDF vs airborne copper and HCl emissions.
Numbers next to data points correspond to run numbers.
PCDD/ PCDF emissions and copper or HCl emissions are
not possible since the limited number of data points
precluded rigorous statistical analyses.
Past studies on MWCs and HWIs have not been able to
prove a causal relationship between chlorine input (and
subsequently HCl emissions) and PCDD/ PCDF emissions
(
21) in Figure 5. Although, most of these selected MWCs
possess various types of air pollution control devices
electrostatic precipitators (ESP), spray dryers, and fabric
[
filters], an emissions factor for source test data measured at
the inlet to the air pollution control device, essentially
uncontrolled, also is presented (21).
(
14), because other more important parameters (e.g., flue
The MWC emission factors used for this comparison were
developed from a compilation of data published on 107
separate test reports (21). These data are averages and are
not representative of any particular facility (21) and are based
on the amount of mass fed into the incinerator. A quality
rating was assigned to each emission factor presented in the
EPA report (21). Specific information on the bases for each
rating for each MWC is in the EPA report (21). In general, the
MWC emission factor data presented in Figure 5 were
assigned high to average ratings (21). The emission factor
rating for the RDF combustor possessing a spray dryer and
fabric filter was assigned a poor rating (21).
gas cleaning temperature, particulate matter carryover,
effectiveness of combustion) completely masked any con-
tribution from chlorine input rates. To heterogeneously form
PCDDs/ PCDFs, a chlorine source, a carbon source, and a
solid-phase metal catalyst must be present. It may be that,
for MWC or HWI facilities, chlorine is present in excess even
at very low concentrations, while in an open-burning
scenario, the known precursor organics and/ or metal cata-
lysts may be present in excess.
Com parison to Other Burn Barrel Study Results. In
general, the PCDD/ PCDF estimated emissions presented in
this paper are higher than the estimated emissions presented
in the Minnesota and Illinois studies but not unreasonably
higher given the variance between test/ fuel conditions
observed in duplicate test burns in this study. The Minnesota
Figure 5 shows that burn barrel PCDD/ PCDF emissions
are generally much higher than for MWCs that possess air
pollution control equipment. Additionally, this figure shows
FIGURE 5. Comparison of total dioxin and furan emissions for burn barrels and municipal waste combustors (source of MWC data: Locating
and Estimating Air Emissions from Sources of Dioxins and Furans; U.S. EPA: Washington, DC, May 1997; EPA-454/R-97-003).
3
8 2
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 3, 2000

modern, well-controlled RDF incinerator emitting PCDDs/
PCDFs at the rate described in Table 6. To perform this
comparison and to be consistent with the MWC emission
rate, the burn barrel estimated emissions were adjusted to
account for the refuse fed into the barrel. It should be noted
that this size MWC facility processes the equivalent waste
from 37 000 nonrecycling and 121 000 avid recycling house-
holds (based on the above waste generation rates). By dividing
the daily estimated emissions from a MWC by the daily
estimated emissions from open burning, it is possible to
estimate how many open-burning households it would take
to equal the amount of PCDDs/ PCDFs emitted by this
moderately sized MWC facility. The number is surprisingly
low, with estimates ranging between 2.5 and 37 households
burning their trash in barrels producing PCDDs/ PCDFs
comparable to a well-operated full-scale RDF MWC facility
serving thousands of households.
TABLE 6. Comparison between Open Burning of Household
Waste and Controlled Combustion of Municipal Waste in a
a
Municipal Waste Combustor (22)
avid
non
recycler recycler
MWC
PCDD (µg/kg of waste burned)
PCDF (µg/kg of waste burned)
PCDD + PCDF (µg/kg of waste
burned)
40.7
222.9
263.6
ndb
6.05
6.05
0.0016
0.0019
0.0035
a
Note that OCDD data were not included in the open-burning results.
nd, none detected.
b
TABLE 7. Number of Open-Burning Households to Equal the
Total PCDD/PCDF Emissions (mg/day) from a Full-Scale MWC
a,b
Facility
Significance of Study and Future Work. This particular
source could potentially be significant in the overall national
PCDD/ PCDF budget (23). The EPA 1994 Draft Dioxin
Reassessment document (24) attempted to conduct a mass
balance for dioxin emissions in the United States and
identified a significant gap between current deposition
estimates and emission estimates. The deposition estimates
were considerably higher than the emissions estimates. The
EPA speculated that this indicated that there were unknown
dioxin emission sources. Given the prevalence of open
burning in barrels (the Illinois study included a survey that
indicated that 40-50% of rural Illinois residents burn at least
a portion of their household waste in barrels), it is very
possible that the PCDD/ PCDF emissions from burn barrels
may be an important missing link to help close the gap
between measured deposition rates and the emissions
inventories.
avid recycler
nonrecycler
PCDD + PCDFc
2.5
37
a
Using refuse generation rate supplied by NYSDEC; MWC burns
82 000 kg/day (200 ton/day). b Em issions data for a m odern, clean-
operating MWC with good com bustion and flue gas cleaning technology.
1
c
OCDD m easurem ents not included in total of PCDDs.
that burn barrel PCDD/ PCDF emissions are higher than for
the mass burn MWC with uncontrolled emissions (i.e.,
emissions measured prior to the flue gas cleaning system).
The emissions for burn barrels were also compared to a
modern, clean-operating MWC with good combustion and
flue gas cleaning technology (22). On the basis of data from
a field test at a RDF MWC (22) and averaging the “Normal
Good” performance tests (PT-08, PT-09, and PT-11) condi-
tions, using the samples taken at the pollution control device
Based on the complex set of variables that affect emissions
of pollutants from burn barrels, further study would be
warranted. The study reported in this paper consisted of a
limited number of tests measuring a wide range of pollutants.
Further studies should concentrate on examining the emis-
sions of PCDDs/ PCDFs. The effect of waste composition,
chlorine content, chlorine type (organic vs inorganic), heating
value, bulk density, and their subsequent effect on resulting
combustion conditions (e.g., temperature) and emissions of
pollutants would be primary variables to investigate. Ad-
ditionally, the types of wastes that are typically disposed of
via open burning in a burn barrel and the frequency of
burning should be investigated.
(
a spray dryer and fabric filter) outlet, the comparison in
Table 6 was generated. Note that the PCDD/ PCDF emission
factor for this particular RDF facility is approximately 3.5
times lower than the average emission factor for this type of
facility presented in Figure 5. Additionally, this emission factor
was based on the mass of refuse fed into the incinerator; an
adjustment to account for the amount of mass burned for
each performance test was not possible. Such an adjustment
would raise the emission factor by an undetermined amount.
This particular MWC study (22) was used for comparison
in the EPA’s burn barrel study report because it contained
a large set of target compounds over a wide range of operating
conditions. Additionally, target compounds from this full-
scale MWC study included not only PCDDs/ PCDFs but also
many of the other compounds (e.g., PCBs, PAHs, etc.) that
were measured in the EPA burn barrel study.
From these comparisons, it is readily apparent that even
the differences between the avid recycler and nonrecycler
emissions are minor in comparison to the difference between
open burning of household waste and the controlled
combustion of municipal waste at most kinds of dedicated
MWC facilities. The emissions from open burning are several
orders of magnitude higher than for controlled combustion
in a modern, clean-operating MWC.
As an additional comparison of a burn barrel and a well-
controlled MWC, an estimate of the number of open-burning
households that emit PCDDs/ PCDFs in amounts comparable
to a hypothetical RDF incinerator was derived. Table 7 was
generated by calculating the total mass of PCDDs/ PCDFs
emitted per day using the estimated emissions per unit mass
burned from Table 6, the waste generation rates derived from
the NYS analysis (4.9 and 1.5 kg/ day for a nonrecycling and
an avid recycling family of four, respectively), assuming a
household burns their waste daily, and comparing those
values to a hypothetical 182 000 kg/ day (200 ton/ day)
Acknowledgments
The authors acknowledge the contributions of Matthew
Pavlik, Peter Kariher, Chris Pressley, Jeff Quinto, Ann Preston,
Jarek Karwowski, Mike Bowling, and Mitch Howell of
ARCADIS Geraghty & Miller. Jeff Ryan also contributed to
this project both with ARCADIS Geraghty & Miller and later
with the U.S. EPA. We would also like to acknowledge the
contributions of Ben Pierson of the New York State Depart-
ment of Environmental Conservation, Division of Solid and
Hazardous Waste, Bureau of Waste Reduction and Recycling,
who provided information about the components of waste
for the avid recycler and the nonrecycler. We would also like
to acknowledge the assistance of Haider Khwaja of WCL&R,
who performed the metals analyses, and of Steve Connor,
also of WCL&R, who performed the dioxin analyses.
Literature Cited
(
1) Emission Characteristics of Burn Barrels; Two Rivers Regional
Council of Public Officials and Patrick Engineering Inc.; Report
prepared for U.S. Environmental Protection Agency Region 5,
June 1994.
VOL. 34, NO. 3, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3 8 3

(
(
(
2) Burn Barrel Dioxin Test; Western Lake Superior Sanitary
District: August 1992.
3) Gerstle, R. W.; Kemnitz, D. A. J. Air Pollut. Control Assoc. 1967,
(18) Winberry, W. T.; Murphy, N. T.; Riggan, R. M. Compendium
Method TO-9: Method for the Determination of Polychlorinated
Dibenzo-p-Dioxins (PCDDs) in Ambient Air Using High-
Resolution Gas Chromatography/ High-Resolution Mass Spec-
trometry (HRGC/ HRMS). Compendium of Methods for the
Determination of Toxic Organic Compounds in Ambient Air;
EPA 600/ 4-89-017 (NTIS PB90-127374); Atmospheric Research
and Exposure Assessment Laboratory, U.S. EPA: Washington,
DC, June 1988.
(19) Determination of Polychlorinated Dibenzo-p-dioxins and Poly-
chlorinated Dibenzofurans from Stationary Sources, EPA Test
Method 23; Code of Federal Regulations, Title 40, Part 60,
Appendix A; U.S. Government Printing Office: Washington, DC,
July 1991.
1
7 (5), 324-327.
4) Lemieux, P. M. Evaluation of Emissions from the Open Burning
of Household Waste in Barrels, Vol. 1; Technical Report EPA-
6
00/ R-97-134a (NTIS PB98-127343); U.S. Government Printing
Office: Washington, DC, November 1997.
(
(
(
5) Olie, K.; Vermeulen, P. L.; Hutzinger, O. Chemosphere 1977, 6,
4
55.
6) Gullett, B. K.; Bruce, K. R.; Beach, L. O. Chemosphere 1990, 20,
945.
7) Gullett, B. K.; Bruce, K. R.; Beach, L. O. Chemosphere 1992, 25,
387.
1
1
(
(
8) Vogg, H.; Stieglitz, L. Chemosphere 1986, 15, 1373.
9) Deacon, H. W. British Patent 1403/ 1863; U.S. Patent 85370/
(
(
(
20) The Analysis of Polychlorinated Dibenzo-p-dioxin and Poly-
chlorinated Dibenzofurans, EPA Method 8280. Test Methods
for Evaluating Solid Waste; SW-846 (NTIS PB87-120291); U.S.
EPA: Washington, DC: September 1986.
21) Locating and Estimating Air Emissions from Sources of Dioxins
and Furans; Office of Air Quality Planning and Standards, U.S.
EPA: Research Triangle Park, NC, May 1997; EPA-454/ R-97-
1
868; U.S. Patent 141.33.
(
(
10) Shaub, W. M.; Tsang, W. Environ. Sci. Technol. 1983, 17, 721.
11) Altwicker, E. R.; Schonberg, J. S.; Konduri, R. K.; Milligan, M. S.
Hazard. Waste Hazard. Mater. 1990, 7 (1).
(
12) Gullett, B. K.; Lemieux, P. M.; Dunn, J. E. Environ. Sci. Technol.
1
994, 28, 107.
0
03.
(
(
13) Kilgroe, J. D. J. Hazard. Mater. 1996, 47, 163-194.
14) Rigo, G. H.; Chandler, A. J.; Lanier, W. S. The Relationship Between
Chlorine in Waste Streams and Dioxin Emissions from Waste
Combustor Stacks; ASME Research Report CRTD-Vol. 36; 1996.
15) Determination of Particulate and Gaseous Mercury Emissions
from Sewage Sludge Incinerators; 40 CFR, Part 61, Appendix B,
Method 101A; Revised as of July 1, 1991; U.S. Government
Printing Office: Washington, DC, 1991.
22) Finkelstein, A.; Klicius, R. D. National Incinerator Testing and
Evaluation Program: The Environmental Characterization of
Refuse-derived Fuel (RDF) Combustion Technology, Mid-Con-
necticut Facility, Hartford, Connecticut; EPA-600/ R-94-140 (NTIS
PB96-153432); December 1994.
(
(
23) Hileman, B. Chem. Eng. News 1998, June 29.
(24) Public Draft Report: Estimating Exposure to Dioxin-Like
Compounds; Office of Health and Environmental Assessment,
U.S. EPA: Washington, DC, June 1994.
(
(
16) Determination of Hydrogen Chloride Emissions from Stationary
Sources; 40 CFR Part 60, Appendix A, Method 26; U.S. Govern-
ment Printing Office: Washington, DC, 1994.
17) Reference Method for the Determination of PM10 in the Atmo-
sphere; Title 40, Code of Federal Regulations, Parts 1-51, Part
Received for review April 26, 1999. Revised manuscript re-
ceived October 18, 1999. Accepted November 8, 1999.
5
0, Appendix J; Revised as of July 1, 1993; Office of the Federal
Register; National Archives and Records Administration: Wash-
ington, DC.
ES990465T
3
8 4
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 3, 2000