Paper to be presented at the International Conference on Incineration and Thermal Treatment
Technologies, New Orleans, LA, May 13-17, 2002
A Pilot-Scale Study of the Precursors Leading to the Formation of Mixed Bromo-Chloro
Dioxins and Furans
P.M. Lcmieux and E.S. Stewart.
U.S. Environmental Protection Agency
Office of Research and Development
National Risk Management Research Laboratory
Air Pollution Prevention and Control Division
Research Triangle Park, NC 27711
ABSTRACT
Experiments were performed in a pilot-scale rotary kiln incinerator simulator, where a
mixture of chlorinated and brominated surrogate waste was burned in the presence of injected
flyash from a coal-fired utility boiler. Measurements were made of semivolatile products of
incomplete combustion (PICs), polychlorinated dibenzo-p-dioxins and polychiorinated
dibenzofurans (PCDDs/Fs), and mixed bromo-chloro dibenzo-p-dioxins and furans (PXDDs/Fs).
A statistical analysis of the data has been performed so that variability in the PCDDs/Fs can be
accounted for by variation in the semivolatile PICs, particularly the chlorobcnzcnes (CBz) and
chlorophenols (CPh). In addition, a statistical analysis was performed to investigate the
variability of the PXDDs/Fs as a function of the concentrations of the semivolatile chlorinated,
brominated, and mixed bromo-chloro organics.
INTRODUCTION
Concern from both the public and the U.S. Environmental Protection Agency (EPA) has been
directed at the formation and control of polychlorinated dibenzo-p-dioxins and polychlorinated
dibenzofurans (PCDDs/Fs) from combustion systems. These classes of chemical compounds are
believed to be carcinogenic and have been implicated in other potential health effects, such as
endocrine disruption [1], Br, although not nearly as ubiquitous as CI in combustion systems,
undergoes a similar set of reactions as CI, and has the potential to form a similar class of
compounds, polybrominated dibenzo-p-dioxins and dibenzofurans (PBDDs/Fs). There are an
additional 5020 congeners of the mixed bromo-chloro dioxins and furans; PXDDs/Fs (X = CI,
Br). The toxicity of PBDDs/Fs is comparable to that of their chlorinated analogs [2,3].
However, there are only a limited amount of data on emissions of brominated compounds from
incinerators, and most of the data in the literature were generated in laboratory experiments [4],
As more and more discarded consumer electronics equipment, which frequently contains
brominated flame retardants, enters the waste stream, understanding the impact of the presence
of Br on combustion emissions becomes important.
The U.S. EPA has performed pilot-scale investigations into the interaction between Br and CI
and its influence on emissions of volatile organic PICs and PCDDs/Fs [5-9]. Br has been shown
to influence emissions of chlorinated organics; however, the phenomenon appears to be sensitive
to temperature and local stoichiometry (e.g., mixing). Recent experiments have observed that
the formation of PXDDs/Fs relative to the formation of PCDDs/Fs is a function of the Br/CI
molar ratio across a range of combustion conditions. Since previous work [9] has observed that
1
-------
the PXDDs/Fs were mostly a function of the Br/CI molar ratio, it suggests that, in the context of
precursor-based formation of PXDDs/Fs (as opposed to de novo synthesis), the bromo-chloro
precursors to the PXDDs/Fs would also be a function of the Br/CI molar ratio, and it might be
possible that the non-brominated precursors to PCDDs/Fs may be reasonable surrogates for the
emissions of PCDDs/Fs independent of the Br/CI molar ratio. The purpose of this paper is to
examine whether variability in the PCDDs/Fs can be explained using the semivolatile surrogates
of CBz and CPh over the entire range of combustion conditions and Br/CI molar ratios
previously studied. Another purpose of this paper is to examine which precursors can account
for the formation of the PXDDs/Fs. Successful development of surrogate performance indicators
that are useful across a wide cross section of fuel compositions would be useful for both industry
and the regulatory community.
EXPERIMENTAL
Experiments were performed on EPA's Rotary Kiln Incinerator Simulator (RKIS), a 73-kW
(250,000-Btu/hr) pilot-scale simulator with a 73-kW secondary combustion chamber (SCC). The
RKIS is shown in Figure 1. After exiting the SCC, the flue gases pass through a long horizontal
duct, where organic species are sampled. Gases then pass through a flue gas cleaning system
(FGCS) consisting of an afterburner, spray quench, baghouse, and wet scrubber. The long
horizontal duct provides several seconds of gas-phase residence time during which the PCDD/F
formation reactions can occur. During these experiments the temperature profile in the duct was
maintained at a constant value ranging from 369 °C at point Duct 3 to 296 °C at point Duct 6.
Note that the Duct 7 thermocouple (TC) shown in Figure 1 is not on the same run of ducting as
Ducts 5 and 6 TCs, but on another vertical run leading to the FGCS manifold.
Duct 2. Duct 3 Duct 4 FGCS
Figure 1. Rotary Kiln Incinerator Simulator
2
-------
The RKIS was fired with natural gas in the primary burner, and a mixture of natural gas and
surrogate waste in the SCC. Liquid surrogate waste mixtures of methylene chloride (CH2CI2,
MeCb) and methylene bromide (CH2Br2, MeBr2) were atomized directly into the afterburner
(along with the burner's natural gas fuel) as shown in Figure 1. Air flowing at a rate of 0.42
scm/h (15 scfh) was used to atomize 25 mL/min of the liquid.
Reactive surfaces for the heterogeneous reactions that form PCDDs/Fs were provided by
injecting 1 g/min of flyash collected from a coal-fired utility boiler into the transition section
between the kiln and the SCC using a K-Tron gravimetric screw feeder and an air eductor. This
resulted in gas-phase particulate matter (PM) loadings of approximately 35 mg/dscm, which is
near the upper PM loading allowable for facilities operating under a Resource Conservation and
Recovery Act (RCRA) permit. The flyash was analyzed for catalytic metals and other key
components, and was found to have the following concentrations, all in |u.g/g: copper (Cu): 169;
aluminum (Al): 92,800; silicon (Si): 256,000; and iron (Fe): 41,500.
Continuous emission monitors (CEMs) provided simultaneous measurement of oxygen (O2),
carbon dioxide (CO2), carbon monoxide (CO), and nitric oxide (NO) concentrations at two
locations. A Perkin-Elmer MCS100 measured concentrations of water (H2O) and hydrogen
chloride (HC1) in the duct, while providing a backup CO2 measurement.
Sampling, analysis, and combustor operation were performed as described earlier [9].
Chlorobenzenes and chlorophenols were analyzed from Method 23 extracts using GC/MS
screening techniques; however, since no CBz or CPh presampling surrogates were used,
recoveries were not estimated. Table 1 lists the designated test conditions.
Table 1. Test conditions
Condition
Kiln Air
Kiln Gas
SCC Air
SCC Gas
(scm/h)
(scm/h)
(scm/h)
(scm/h)
0 (Baseline)
113
5.7
42
4.2
1
57
4.2
42
4.2
2
57
4.2
51
5.1
3
113
4.2
42
4.2
4
57
5.7
42
4.2
5
57
5.7
51
5.1
6
113
5.7
51
5.1
7
113
4.2
51
5.1
8
113
5.7
54
4.2
RESULTS
The chlorobenzene (CBz) and chlorophenol (CPh) portions of the semivolatile organic
compound (SVOC) data were converted to units of |j.g/dscm corrected to 12% CO2. These data
are shown in Table 2. The PCDD/F data were converted to ng/dscm corrected to 12% CO2. The
PCDD/F and PXDD/F data are shown in Table 3. The data were imported into the SAS JMP®
software, and the run on 2/26/01 was excluded due to its being previously determined [9] to be
an outlier. For the purposes of this analysis, non-detects were set to zero.
3
-------
Table 2, Chlorobenzene, bromochlorobenzene, and chlorophenol data ()ig/dscm corrected to 12% CO2)
a
3
b:
c
0
-0
c
0
0
CO
QJ
H
_g
"3
3.
0
5
u
m
c
N
c
_o
0
_e
U
'a
G
ss
N
C
WO
O
s
2
0
5
0
c
N
f
4i
£1
g
O
"u
Q
iD
C
O
N
a
V
.0
0
8
5
CN
c
N
a
a
x>
0
s
0
H
eT
V
s
a>
tz
&
a
5
3
H
r-4
£
4>
&
XI
2
0
"S
CN
O
c
«
N
e:
4>
O
O
'A *0
t-; g
CN 0
~ h-
3
s
5
JO
0
0
(N t>
-"H
«
C
to
N
C
£s
O
3
^ "S
^ 2
CN £
__r (—
a>
c
o
N"
K
¦s
0
2
K
-------
Table 3. PCDD/F and PXDD/F Data (corrected to 12% C02)
Pi
B
o
CO
-a
~Bb
Q
Q
U
CL|
a
o
Hi
Q
U
On
Pi
Q
O
PL,
Q If
Q °
Fn 01
O T3
fc/ ~Sb
Wfl
fi
0
xn
T3
a
w
H
1
M
W)
w
Q
a
w
H
a
o
tn
a
w
H
60
a
a
w
H
a
o
m
T3
a
w
H
60
u,
D
a
w
H
Ph
Q
•«
CQ
O
x
ffi
a
o
bO
Ph
Q
i-i
m
Q
0
o
eu
a
o
bO
Pi
D
t-i
m
'C
H
U
(D
H
a
o
bfl
Pn
Q
CO
o
Oh
a
a
o
C/3
~5b
Pi
Q
Vh
CQ
Q
O
x
ffi
6/9/00
12.44
263.2
275.6
0.06
1.85
1.91
NM
NM
NM
NM
NM
7/13/00
22.24
383.5
405.8
0.06
2.61
2.67
NM
NM
NM
NM
NM
7/14/00
10.48
184.6
195.1
0.03
1.87
1.89
NM
NM
NM
NM
NM
7/27/00
15.67
238.0
253.7
0.04
2.15
2.19
NM
NM
NM
NM
NM
8/3/00
9.83
139.8
149.6
0.03
1.20
1.24
NM
NM
NM
NM
NM
8/4/00
14.10
236.2
250.3
0.04
1.64
1.68
NM
NM
NM
NM
NM
8/8/00
13.89
133.0
146.8
0.03
1.48
1.51
NM
NM
NM
NM
NM
8/9/00
7.20
105.9
113.1
0.02
1.30
1.32
NM
NM
NM
NM
NM
8/16/00
11.58
195.6
207.2
0.04
1.93
1.97
NM
NM
NM
NM
NM
8/18/00
18.20
219.7
237.9
0.07
2.69
2.77
NM
NM
NM
NM
NM
2/28/01
16.50
213.6
230.1
0.05
1.90
1.95
11,8
1.8
ND
7.3
ND
3/7/01
12.65
141.5
154.2
0.04
1.49
1.53
18.8
3.0
ND
9.9
ND
3/14/01
ND
55.3
55.3
ND
0.58
0.58
9.1
ND
ND
4.1
ND
3/16/01
6.54
85.7
92.2
0.02
1.12
1.13
9.2
2.1
ND
5.7
0.7
3/21/01
9.46
83.3
92.8
0.02
0.91
0.93
12.5
3.8
ND
7.5
ND
4/11/01
2.84
28.3
31.1
0.00
0.32
0.32
14.2
11.8
ND
7.1
1.2
4/13/01
11.28
81.1
92.3
0.03
1.21
1.23
12.0
3.4
ND
6.0
ND
4/17/01
6.63
42.3
49.0
0.01
0.52
0.53
10.2
3.8
ND
2.6
ND
4/19/01
7.67
185.6
193.3
0.02
3.25
3.27
NM
NM
NM
NM
NM
5/2/01
8.41
62.8
71.3
0.03
0.87
0.90
NM
NM
NM
NM
NM
5/7/01
4.56
44.5
49.0
ND
0.69
0.69
NM
NM
NM
NM
NM
5/9/01
9.85
160.2
170.0
0.03
2.29
2.32
NM
NM
NM
NM
NM
ND - Not
NM - Not
Detected
Measured
The dataset was subjected to a STEPWISE statistical analysis to determine if individual CBz
or CPh compounds could account for the variations in the PCDD, PCDF, ^TCDD+PCDF, the I-
TEQ contribution from PCDDs (TEQd), the I-TEQ contribution from PCDFs (TEQf), and the
combined I-TEQ contribution from PCDDs/Fs (TEQdf)- Because some of the runs were
performed with no brominated feed and some were performed with brominated feed, the analysis
was repeated 3 times, first with the complete dataset (N=22), then with subsets of the data with
the Br/Cl=0 runs (N=l 1), and the Br/C1^0 (N=l 1) runs. A least squares analysis was then
performed using £CBz and £CPh as the independent (X) variables. Statistical significance was
determined by use of the "P value" which, if P < 0.05, the predictor exhibited a statistically
significant effect.
5
-------
In general, correlations were found for the totals and the TEQs, although the significant
predictors for the totals were usually not the same significant predictor of the TEQs. The R2
values were typically in the 0.3-0.6 range, although the P values showed statistically significant
correlations. The PCDDs in general gave poorer correlations than the PCDFs, at least in part
because the concentrations of the PCDDs were much lower than the concentrations of the PCDFs
and the non-detects therefore had a greater influence on the PCDD concentrations and TEQs.
For the runs where brominated feed was present, PCDDs were extremely low, and analysis of the
variance of the PCDDs and the TEQd yielded no correlations. In general, similar trends were
found in the complete dataset and both the non-brominated and brominated subsets, except
where the concentrations of many of the targets were too low (as was the case for many of the
PCDDs in the brominated subset), so the correlations to be presented will be based on the
complete dataset. Other than the presence of mixed bromochloro furans [9] that was found when
brominated materials were being burned, the concentrations of the purely chlorinated SVOCs
showed the same statistical trends when compared to PCDDs/Fs.
An interesting observation was that the concentration of 2,4,6-trichlorophenol appeared as a
significant predictor of the PCDDs, PCDFs, and the £(PCDDs+PCDFs), whereas various
combinations of chlorobenzenes, particularly hexachlorobenzene, appeared as significant
predictors of TEQd, TEQf, and TEQdf- This observation is illustrated in Figure 2. The left
portion of Figure 2 shows 2,4,6-trichlorophenol vs. £(PCDDs+PCDFs), with the line showing
the least squares fit (R2=0.31). The right portion of Figure 2 shows hexachlorobenzene vs.
TEQdf, with the fitted line reflecting a second order polynomial fit with an R2=0.53. It is likely
that the relatively low R2 in the left portion is due to the one data point on the Y-axis where a
high level of PCDD/F was observed. This could be either an outlier due to analytical limitations
or support for the fact that the PCDD/F formation mechanism is highly complex, and that other
aspects of the formation mechanism (e.g., gas-phase formation, de novo synthesis) might be
influencing the PCDD/F emissions.
CO
E
~o>
c:
"oT
UL
Q
O
Q.
+
to
Q
Q
O
CL
l/sl
450-
400-1
350
300
250
200
150
100-1
¦
2 •
R =0.53
R2=0.31
1
©/ *
V %/
J' ¦¦
f
¦ ¦
%. J
-------
Examining the ability of ]TCBz and £CPh to account for the variability in the PCDDs/Fs
yielded similar results to the individual isomers of CBz and CPh, although some of the
correlations were slightly improved in terms of the R2 parameter. Figure 3 plots the £CPh vs.
the TEQdf-
ICPh (f/g/m3)
Figure 3. ICPh vs. PCDDs/Fs TEQs
The only PXDFs that were detected in all of the samples where MeBr2 was fed were the
hexachloromonobromo dibenzofuran and heptachloromonobromo dibenzofuran. Other
identified PXDFs were not commonly detected in many of the samples, which made it difficult
to do a statistical analysis of their variability. Three mixed bromo-chlorobenzenes were
identified in the samples. A least squares analysis was performed using each of the
bromoehlorobenzenes. It yielded a 1 parameter model with trichlorotribromobenzene with an
R2=0.59 with HxClMBrDF and an R2=0.61 with HpClMBrDF. These trends are illustrated in
Figure 4.
Trichlorotribromobenzene Cyg/m3)
Figure 4. Trichlorotribromobenzene vs. PXDFs
7
-------
CONCLUSIONS
A series of experiments were performed in a pilot-scale rotary kiln incinerator simulator
where a surrogate waste feed consisting of mixtures of methylene chloride and methylene
bromide were combusted with natural gas. The combustor was operated over a range of
temperatures, oxygen concentrations, and Br/CI molar ratios. Chlorobenzenes, chlorophenols,
and bromochlorobenzenes were measured. A statistical data analysis was performed to examine
whether relationships existed between those compounds and the PCDDs/Fs (both totals and I-
TEQs) and PXDFs that were measured. The following observations were made:
Some of the individual isomers of the CBz and CPh gave statistically significant correlations
with PCDDs/Fs and the I-TEQs. The 2,4,6-trichlorophenol gave the best fit for the
£(PCDD+PCDF), and hexachlorobenzene gave the best fit for TEQdf.
The £CPh was a statistically significant predictor for TEQdf-
Trichlorotribromobenzene was a statistically significant predictor of the PXDFs that were
identified.
The PCDDs/Fs and PXDFs that were identified in these tests were all highly halogenated.
The precursors that gave the best fits were, for the most part, also highly halogenated. This is
consistent with a formation path from methylene bromide and methylene chloride to halogenated
aromatics, then additional halogenation steps to produce the perhalogenated aromatics, followed
by reactions to form perhalogenated, or nearly perhalogenated, dioxins and furans.
REFERENCES
1. Okey, A.B., D. Riddick, and P. Harper, "The Ah Receptor: Mediator of the Toxicity of
2,3,7,8-TCDD and Related Compounds," Toxicology Letters, 70: 1-22, 1994.
2. Nagao, T., D. Neubert, and E. Loser, "Comparative Studies on the Induction of Ethoxy
Resorufin o-deethylase by 2,3,7,8-TCDD and 2,3,7,8-TBrDD," Chemosphere, 20, 1189-
1192, 1990.
3. Mermear, J.H. and C.C. Lee, "Polybrominated Dibenzo-p-Dioxins and Dibenzofurans:
Literature Review and Health Assessment," Environmental Health Perspectives
Supplements, Vol. 102 Suppl 1:265-274, 1994.
4. Wellington Laboratories, Inc., "Polybromodibenzo-p-dioxins and Furans, Their Mixed
Bromo/Chloro Analogues: A Review of the Literature on Sources, Methods of Analysis,
Environmental Distribution, and Toxicities," Ontario Ministry of the Environment Report
96- 02403, April 1984, Toronto, Ontario, Canada.
5. Ryan, J.V., P.M. Lemieux, C. Lutes, and D. Tabor, "Development of PIC Target Analyte
List for Hazardous Waste Incineration Processes," in Proceedings of International
Conference on Incineration and Thermal Treatment Technologies, Savannah, GA, May 6-
10, 1996.
6. Lemieux, P.M., J.V. Ryan, C. Lutes, and K. Bruce, "Interactions Between Bromine and
Chlorine in a Pilot-Scale Hazardous Waste Incinerator," in Proceedings of International
Conference on Incineration and Thermal Treatment Technologies, Savannah, GA, May 6-
10, 1996.
8
-------
f
7. Lemieux, P.M. and J.V. Ryan, "Enhanced Formation of Chlorinated PICs by the Addition
of Bromine," Combust. Sci, and Tech., Vol. 134, 1-6, pp. 367-387, 1998.
8. Lemieux, P.M. and J.V. Ryan, "Enhanced Formation of Dioxins and Furans from
Combustion Devices by Addition of Trace Quantities of Bromine," Waste Management,
Vol. 18, pp. 361-370, 1998.
9. Lemieux, P.M., E.S. Stewart, and J.V. Ryan, "Pilot-Scale Studies on the Effect of Bromine
Addition on the Emissions of Chlorinated Organic Combustion By-Products," Waste
Management, in press 2002.
-------
n™ t, rnn TECHNICAL REPORT DATA „„„„„
NRMRL-RTP-P-b / / (Please read Instructions on the reverse before completing PB200
2-107515
IIIIHIflll
1. REPORT NO. 2.
EPA/600/A-02/074
3RE 111
4. TITLE AND SUBTITLE
A Pilot-scale Study of the Precursors Leading to the
Formation of Mixed Bromo-chloro Dioxins and Furans
5. REPORT DATE
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Paul M. Lemieux and Eric S. Stewart
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
See Block 12
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
68-C-99-201, W.A. 2-001
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Air Pollution Prevention and Control Division
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Published paper; 6/00-12/01
14. SPONSORING AGENCY CODE
EPA/600/13
15. supplementary notes ^ppCD project officer is Paul M. Lemieux, E305-01, 919/541-0962.
For International Conference on Incineration and Thermal Treatment Technologies,
New Orleans, LA, 5/14/02,
i6. abstract paper gives results of experiments in a pilot-scale rotary kiln incinera-
tor simulator, where a mixture of chlorinated and brominated surrogate waste was
burned in the presence of injected fly-ash from a coal-fired utility boiler. Measure-
ments were made of semivolatile products of incomplete combustion (PICs), polychlor-
inated aibenzo-p-dioxins and polychlorinated dibenzofurans (PCDDs/Fs) and mixed
bromo-chloro dibenzo-p-dioxins and furans (PXDDs/Fs). The data have been analyzed
statistically so that variability in the PCDDs/Fs can be accounted for by variation in
the semivolatile PICs, particularly the chlorobenzenes (CBz) and chlorophenols (CPh).
In addition, a statistical analysis was performed to investigate the variability of
the PXDDs/Fs as a function of the concentrations of the semivolatile chlorinated,
brominated, and mixed bromo-chloro organics.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Air Pollution Furans
Coal Bromine
Combustion Chlorine
Utilities Wastes
Boilers
Fly Ash
Halohyarocarbons
Stationary Sources
Dioxins
13B
21D 07B
21B
14G
14A
07C
18. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
9
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
-------
------- |